{
  "status": "BOTH_FINITE_RUNS_CLOSED_AND_COLLECTED_WITH_DISCLOSED_CENSORING",
  "utc": "2026-10-27T03:33:32.796217+00:00",
  "models": {
    "flash": {
      "model": {
        "tag": "flash",
        "repo": "deepseek-ai/DeepSeek-V4.1-Flash-1017",
        "revision": "1bae0e1088993b9b82de08f59de5be5a34e42170",
        "profiles": {
          "short_science_qa": {
            "max_tokens": 262144,
            "timeout_s": 5400,
            "chat_template_kwargs": {
              "thinking": true,
              "reasoning_effort": 25
            },
            "temperature": 1.0,
            "top_p": 0.95
          },
          "long_document_grounded_qa": {
            "max_tokens": 262144,
            "timeout_s": 5400,
            "chat_template_kwargs": {
              "thinking": true,
              "reasoning_effort": 25
            },
            "temperature": 1.0,
            "top_p": 0.95
          },
          "scientific_reasoning": {
            "max_tokens": 262144,
            "timeout_s": 5400,
            "chat_template_kwargs": {
              "thinking": true,
              "reasoning_effort": 100
            },
            "temperature": 1.0,
            "top_p": 0.95
          },
          "scientific_code": {
            "max_tokens": 262144,
            "timeout_s": 5400,
            "chat_template_kwargs": {
              "thinking": true,
              "reasoning_effort": 100
            },
            "temperature": 1.0,
            "top_p": 0.95
          },
          "gpqa_diamond": {
            "max_tokens": 262144,
            "timeout_s": 5400,
            "chat_template_kwargs": {
              "thinking": true,
              "reasoning_effort": 100
            },
            "temperature": 1.0,
            "top_p": 0.95
          }
        }
      },
      "context_tokens": 1048576,
      "max_output_tokens": 262144,
      "gpu_memory_fraction": 0.8,
      "gpqa": {
        "condition": "flash-gpqa_diamond",
        "concurrency": 20,
        "planned": 198,
        "started": 198,
        "complete": 198,
        "statuses": {
          "complete": 198
        },
        "automatic_pass": 184,
        "automatic_fail": 14,
        "rubric_only": 0,
        "unscored_incomplete": 0,
        "latency_s_complete_only": {
          "first_generated_s": {
            "n": 198,
            "median": 0.07878894748864695,
            "p95_nearest_rank": 1.3921910559292883,
            "min": 0.0711213800823316,
            "max": 3.4673416489968076
          },
          "first_final_s": {
            "n": 198,
            "median": 57.57311679451959,
            "p95_nearest_rank": 359.4964162019314,
            "min": 0.4236030150204897,
            "max": 737.7693776750239
          },
          "completion_s": {
            "n": 198,
            "median": 58.4063227045699,
            "p95_nearest_rank": 359.5188735569827,
            "min": 0.9449617450591177,
            "max": 737.790664169006
          }
        },
        "elapsed_s": 1113.7984421469737,
        "actual_prompt_tokens": {
          "n": 198,
          "median": 249.0,
          "p95_nearest_rank": 491,
          "min": 121,
          "max": 2602
        },
        "actual_completion_tokens": {
          "n": 198,
          "median": 8354.0,
          "p95_nearest_rank": 51188,
          "min": 125,
          "max": 106069
        },
        "effective_timeout_s": {
          "n": 198,
          "median": 5400.0,
          "p95_nearest_rank": 5400,
          "min": 5400,
          "max": 5400
        },
        "deadline_censored": 0,
        "requests_sha256": "d57b60c78f87cd774aef265d6fee65b6edef19362c4c8ad81affbfc845a7cd29",
        "profile": {
          "max_tokens": 262144,
          "timeout_s": 5400,
          "chat_template_kwargs": {
            "thinking": true,
            "reasoning_effort": 100
          },
          "temperature": 1.0,
          "top_p": 0.95
        },
        "correct_fraction_of_all_198": 0.9292929292929293,
        "correct_fraction_of_completed": 0.9292929292929293
      },
      "science": {
        "condition": "flash-scientific_reasoning",
        "concurrency": 5,
        "planned": 20,
        "started": 20,
        "complete": 20,
        "statuses": {
          "complete": 20
        },
        "automatic_pass": 11,
        "automatic_fail": 1,
        "rubric_only": 8,
        "unscored_incomplete": 0,
        "latency_s_complete_only": {
          "first_generated_s": {
            "n": 20,
            "median": 0.07269003352848813,
            "p95_nearest_rank": 0.1548374789999798,
            "min": 0.07029371301177889,
            "max": 0.15527361095882952
          },
          "first_final_s": {
            "n": 20,
            "median": 2.4361192669603042,
            "p95_nearest_rank": 14.288928673951887,
            "min": 0.5454481900669634,
            "max": 19.334483694052324
          },
          "completion_s": {
            "n": 20,
            "median": 3.7976564139826223,
            "p95_nearest_rank": 16.942980961990543,
            "min": 1.4690543920733035,
            "max": 21.82646722404752
          }
        },
        "elapsed_s": 32.8518773909891,
        "actual_prompt_tokens": {
          "n": 20,
          "median": 199.0,
          "p95_nearest_rank": 214,
          "min": 187,
          "max": 218
        },
        "actual_completion_tokens": {
          "n": 20,
          "median": 651.5,
          "p95_nearest_rank": 3286,
          "min": 254,
          "max": 4602
        },
        "effective_timeout_s": {
          "n": 20,
          "median": 5395.227975989459,
          "p95_nearest_rank": 5399.998582943925,
          "min": 5386.553664591978,
          "max": 5399.998829749995
        },
        "deadline_censored": 0,
        "requests_sha256": "dd7fbe1785bedd82f63e3e17f5cba8eff7c7c115553ee3c3a6d63550335be7b7",
        "profile": {
          "max_tokens": 262144,
          "timeout_s": 5400,
          "chat_template_kwargs": {
            "thinking": true,
            "reasoning_effort": 100
          },
          "temperature": 1.0,
          "top_p": 0.95
        }
      },
      "code": {
        "generated": 10,
        "task_pass": 10,
        "task_fail": 0,
        "fixed_tests": 61,
        "test_pass": 61,
        "failed_cases": []
      },
      "operational": [
        {
          "condition": "flash-fair262k-c1-long_document_grounded_qa",
          "concurrency": 1,
          "planned": 20,
          "started": 20,
          "complete": 20,
          "statuses": {
            "complete": 20
          },
          "automatic_pass": 20,
          "automatic_fail": 0,
          "rubric_only": 0,
          "unscored_incomplete": 0,
          "latency_s_complete_only": {
            "first_generated_s": {
              "n": 20,
              "median": 0.33771378255914897,
              "p95_nearest_rank": 0.34186441800557077,
              "min": 0.33577759901527315,
              "max": 1.6251776979770511
            },
            "first_final_s": {
              "n": 20,
              "median": 0.9383142085280269,
              "p95_nearest_rank": 1.1295272540301085,
              "min": 0.7098339980002493,
              "max": 2.3320820349035785
            },
            "completion_s": {
              "n": 20,
              "median": 1.7328429645276628,
              "p95_nearest_rank": 1.8890361099038273,
              "min": 1.4095612959936261,
              "max": 3.0232595559209585
            }
          },
          "elapsed_s": 35.17608584591653,
          "actual_prompt_tokens": {
            "n": 20,
            "median": 17316.0,
            "p95_nearest_rank": 17316,
            "min": 17316,
            "max": 17316
          },
          "actual_completion_tokens": {
            "n": 20,
            "median": 326.0,
            "p95_nearest_rank": 347,
            "min": 251,
            "max": 363
          },
          "effective_timeout_s": {
            "n": 20,
            "median": 1760.406563174096,
            "p95_nearest_rank": 1775.0301000190666,
            "min": 1744.7669108000118,
            "max": 1778.0539727550931
          },
          "deadline_censored": 0,
          "requests_sha256": "298f86282e1c7bf2b8298f57a2da0ccddc2524dacd89eabd4f629b212a96a6d2",
          "profile": {
            "max_tokens": 262144,
            "timeout_s": 5400,
            "chat_template_kwargs": {
              "thinking": true,
              "reasoning_effort": 25
            },
            "temperature": 1.0,
            "top_p": 0.95
          }
        },
        {
          "condition": "flash-fair262k-c1-scientific_reasoning",
          "concurrency": 1,
          "planned": 20,
          "started": 20,
          "complete": 20,
          "statuses": {
            "complete": 20
          },
          "automatic_pass": 12,
          "automatic_fail": 0,
          "rubric_only": 8,
          "unscored_incomplete": 0,
          "latency_s_complete_only": {
            "first_generated_s": {
              "n": 20,
              "median": 0.06759888294618577,
              "p95_nearest_rank": 0.06917750206775963,
              "min": 0.06607994297519326,
              "max": 0.06929693999700248
            },
            "first_final_s": {
              "n": 20,
              "median": 1.7181373700150289,
              "p95_nearest_rank": 20.49449132801965,
              "min": 0.5452149930642918,
              "max": 21.563211795990355
            },
            "completion_s": {
              "n": 20,
              "median": 2.7944583680364303,
              "p95_nearest_rank": 22.859037535032257,
              "min": 1.0758525310084224,
              "max": 23.108468280057423
            }
          },
          "elapsed_s": 114.0878863359103,
          "actual_prompt_tokens": {
            "n": 20,
            "median": 199.0,
            "p95_nearest_rank": 214,
            "min": 187,
            "max": 218
          },
          "actual_completion_tokens": {
            "n": 20,
            "median": 646.0,
            "p95_nearest_rank": 5351,
            "min": 241,
            "max": 5410
          },
          "effective_timeout_s": {
            "n": 20,
            "median": 1716.42840357346,
            "p95_nearest_rank": 1733.6081464539748,
            "min": 1633.6319052609615,
            "max": 1742.8759060830344
          },
          "deadline_censored": 0,
          "requests_sha256": "5cb6e542feac0f1d2fa282d07026b6b848f6456349ebc229f6c75a7ab76cf60f",
          "profile": {
            "max_tokens": 262144,
            "timeout_s": 5400,
            "chat_template_kwargs": {
              "thinking": true,
              "reasoning_effort": 100
            },
            "temperature": 1.0,
            "top_p": 0.95
          }
        },
        {
          "condition": "flash-fair262k-c1-short_science_qa",
          "concurrency": 1,
          "planned": 20,
          "started": 20,
          "complete": 20,
          "statuses": {
            "complete": 20
          },
          "automatic_pass": 20,
          "automatic_fail": 0,
          "rubric_only": 0,
          "unscored_incomplete": 0,
          "latency_s_complete_only": {
            "first_generated_s": {
              "n": 20,
              "median": 0.06640748347854242,
              "p95_nearest_rank": 0.06986434094142169,
              "min": 0.06581807800102979,
              "max": 0.07358841795939952
            },
            "first_final_s": {
              "n": 20,
              "median": 0.40131706197280437,
              "p95_nearest_rank": 0.8619877820601687,
              "min": 0.26341481099370867,
              "max": 0.8620266709476709
            },
            "completion_s": {
              "n": 20,
              "median": 1.035499827528838,
              "p95_nearest_rank": 1.5109517601085827,
              "min": 0.6255984690506011,
              "max": 1.6239429989363998
            }
          },
          "elapsed_s": 21.94371176394634,
          "actual_prompt_tokens": {
            "n": 20,
            "median": 186.0,
            "p95_nearest_rank": 194,
            "min": 173,
            "max": 198
          },
          "actual_completion_tokens": {
            "n": 20,
            "median": 231.0,
            "p95_nearest_rank": 344,
            "min": 134,
            "max": 371
          },
          "effective_timeout_s": {
            "n": 20,
            "median": 1789.3065106485155,
            "p95_nearest_rank": 1799.2871822121087,
            "min": 1778.860854240018,
            "max": 1799.9995055090403
          },
          "deadline_censored": 0,
          "requests_sha256": "5e57d8844dbe0bd8383eecb5cd293c02dcf51d5eb3efca5f74bd76eb3574cd30",
          "profile": {
            "max_tokens": 262144,
            "timeout_s": 5400,
            "chat_template_kwargs": {
              "thinking": true,
              "reasoning_effort": 25
            },
            "temperature": 1.0,
            "top_p": 0.95
          }
        },
        {
          "condition": "flash-fair262k-c10-long_document_grounded_qa",
          "concurrency": 10,
          "planned": 20,
          "started": 20,
          "complete": 20,
          "statuses": {
            "complete": 20
          },
          "automatic_pass": 20,
          "automatic_fail": 0,
          "rubric_only": 0,
          "unscored_incomplete": 0,
          "latency_s_complete_only": {
            "first_generated_s": {
              "n": 20,
              "median": 0.6747568545397371,
              "p95_nearest_rank": 2.6018465739907697,
              "min": 0.36233892501331866,
              "max": 4.123771516024135
            },
            "first_final_s": {
              "n": 20,
              "median": 4.691072714456823,
              "p95_nearest_rank": 5.164662490948103,
              "min": 1.0189679219620302,
              "max": 5.168699292931706
            },
            "completion_s": {
              "n": 20,
              "median": 5.766351913975086,
              "p95_nearest_rank": 9.07819852605462,
              "min": 2.1258481930708513,
              "max": 9.844160073902458
            }
          },
          "elapsed_s": 11.970924291061237,
          "actual_prompt_tokens": {
            "n": 20,
            "median": 17316.0,
            "p95_nearest_rank": 17316,
            "min": 17316,
            "max": 17316
          },
          "actual_completion_tokens": {
            "n": 20,
            "median": 306.0,
            "p95_nearest_rank": 353,
            "min": 257,
            "max": 377
          },
          "effective_timeout_s": {
            "n": 20,
            "median": 1563.0377107505337,
            "p95_nearest_rank": 1565.907587198075,
            "min": 1556.0627994340612,
            "max": 1565.9078863729956
          },
          "deadline_censored": 0,
          "requests_sha256": "b7359293bf3821a48239f78f197e0571df79544ff692dc616cdab02a1a655b6b",
          "profile": {
            "max_tokens": 262144,
            "timeout_s": 5400,
            "chat_template_kwargs": {
              "thinking": true,
              "reasoning_effort": 25
            },
            "temperature": 1.0,
            "top_p": 0.95
          }
        },
        {
          "condition": "flash-fair262k-c10-scientific_reasoning",
          "concurrency": 10,
          "planned": 20,
          "started": 20,
          "complete": 20,
          "statuses": {
            "complete": 20
          },
          "automatic_pass": 12,
          "automatic_fail": 0,
          "rubric_only": 8,
          "unscored_incomplete": 0,
          "latency_s_complete_only": {
            "first_generated_s": {
              "n": 20,
              "median": 0.07481569243827835,
              "p95_nearest_rank": 0.14488247490953654,
              "min": 0.0715688019990921,
              "max": 0.14536134607624263
            },
            "first_final_s": {
              "n": 20,
              "median": 3.119140033959411,
              "p95_nearest_rank": 9.893676184001379,
              "min": 0.7429152929689735,
              "max": 20.01637332199607
            },
            "completion_s": {
              "n": 20,
              "median": 4.737114107992966,
              "p95_nearest_rank": 11.747242024051957,
              "min": 1.717243982013315,
              "max": 23.16619716095738
            }
          },
          "elapsed_s": 26.941651628003456,
          "actual_prompt_tokens": {
            "n": 20,
            "median": 199.0,
            "p95_nearest_rank": 214,
            "min": 187,
            "max": 218
          },
          "actual_completion_tokens": {
            "n": 20,
            "median": 679.5,
            "p95_nearest_rank": 2135,
            "min": 261,
            "max": 4871
          },
          "effective_timeout_s": {
            "n": 20,
            "median": 1553.072998995951,
            "p95_nearest_rank": 1553.9334706809605,
            "min": 1548.6481431309367,
            "max": 1553.933736847015
          },
          "deadline_censored": 0,
          "requests_sha256": "e10232cb3e51eafc905be3c23da8b1fc0dbe2e56cb15af23be587fdecc976dcd",
          "profile": {
            "max_tokens": 262144,
            "timeout_s": 5400,
            "chat_template_kwargs": {
              "thinking": true,
              "reasoning_effort": 100
            },
            "temperature": 1.0,
            "top_p": 0.95
          }
        },
        {
          "condition": "flash-fair262k-c10-short_science_qa",
          "concurrency": 10,
          "planned": 20,
          "started": 20,
          "complete": 20,
          "statuses": {
            "complete": 20
          },
          "automatic_pass": 20,
          "automatic_fail": 0,
          "rubric_only": 0,
          "unscored_incomplete": 0,
          "latency_s_complete_only": {
            "first_generated_s": {
              "n": 20,
              "median": 0.07753541698912159,
              "p95_nearest_rank": 1.4542348550166935,
              "min": 0.07055162906181067,
              "max": 1.4548109759343788
            },
            "first_final_s": {
              "n": 20,
              "median": 1.4241923029185273,
              "p95_nearest_rank": 2.527088513947092,
              "min": 0.45912985398899764,
              "max": 2.574817576096393
            },
            "completion_s": {
              "n": 20,
              "median": 2.2391338250017725,
              "p95_nearest_rank": 3.88343562791124,
              "min": 0.9490037930663675,
              "max": 3.9939204190159217
            }
          },
          "elapsed_s": 5.3975418109912425,
          "actual_prompt_tokens": {
            "n": 20,
            "median": 186.0,
            "p95_nearest_rank": 194,
            "min": 173,
            "max": 198
          },
          "actual_completion_tokens": {
            "n": 20,
            "median": 233.0,
            "p95_nearest_rank": 342,
            "min": 144,
            "max": 344
          },
          "effective_timeout_s": {
            "n": 20,
            "median": 1570.151522817614,
            "p95_nearest_rank": 1571.3084121650318,
            "min": 1567.3129001731286,
            "max": 1571.308664031094
          },
          "deadline_censored": 0,
          "requests_sha256": "91e98c023827cf0a0c910823c2fbc93c6b030b070ee4fc5f5b643ee402bb3e86",
          "profile": {
            "max_tokens": 262144,
            "timeout_s": 5400,
            "chat_template_kwargs": {
              "thinking": true,
              "reasoning_effort": 25
            },
            "temperature": 1.0,
            "top_p": 0.95
          }
        },
        {
          "condition": "flash-fair262k-c20-long_document_grounded_qa",
          "concurrency": 20,
          "planned": 20,
          "started": 20,
          "complete": 20,
          "statuses": {
            "complete": 20
          },
          "automatic_pass": 20,
          "automatic_fail": 0,
          "rubric_only": 0,
          "unscored_incomplete": 0,
          "latency_s_complete_only": {
            "first_generated_s": {
              "n": 20,
              "median": 3.0173747719381936,
              "p95_nearest_rank": 5.391977772931568,
              "min": 0.3704014589311555,
              "max": 5.660321831936017
            },
            "first_final_s": {
              "n": 20,
              "median": 6.311283559014555,
              "p95_nearest_rank": 6.517397202900611,
              "min": 5.725301862927154,
              "max": 6.771681446000002
            },
            "completion_s": {
              "n": 20,
              "median": 7.347275770502165,
              "p95_nearest_rank": 7.622617239947431,
              "min": 6.783165483968332,
              "max": 7.636848711990751
            }
          },
          "elapsed_s": 7.638110750936903,
          "actual_prompt_tokens": {
            "n": 20,
            "median": 17316.0,
            "p95_nearest_rank": 17316,
            "min": 17316,
            "max": 17316
          },
          "actual_completion_tokens": {
            "n": 20,
            "median": 296.5,
            "p95_nearest_rank": 333,
            "min": 234,
            "max": 349
          },
          "effective_timeout_s": {
            "n": 20,
            "median": 1523.232592005108,
            "p95_nearest_rank": 1523.2351566021098,
            "min": 1523.2283587561687,
            "max": 1523.2354737371206
          },
          "deadline_censored": 0,
          "requests_sha256": "dd21232a41bc795a30c05409d951c88d313115336f33f05e1fc6109eed108ac3",
          "profile": {
            "max_tokens": 262144,
            "timeout_s": 5400,
            "chat_template_kwargs": {
              "thinking": true,
              "reasoning_effort": 25
            },
            "temperature": 1.0,
            "top_p": 0.95
          }
        },
        {
          "condition": "flash-fair262k-c20-scientific_reasoning",
          "concurrency": 20,
          "planned": 20,
          "started": 20,
          "complete": 20,
          "statuses": {
            "complete": 20
          },
          "automatic_pass": 12,
          "automatic_fail": 0,
          "rubric_only": 8,
          "unscored_incomplete": 0,
          "latency_s_complete_only": {
            "first_generated_s": {
              "n": 20,
              "median": 1.4762819859897718,
              "p95_nearest_rank": 1.4775254050036892,
              "min": 0.08681353600695729,
              "max": 1.4780471760313958
            },
            "first_final_s": {
              "n": 20,
              "median": 4.462011237046681,
              "p95_nearest_rank": 22.15388894197531,
              "min": 2.0859039400238544,
              "max": 22.203526912024245
            },
            "completion_s": {
              "n": 20,
              "median": 5.971965224016458,
              "p95_nearest_rank": 23.91087432904169,
              "min": 3.0889872680418193,
              "max": 25.605716536054388
            }
          },
          "elapsed_s": 25.60904584999662,
          "actual_prompt_tokens": {
            "n": 20,
            "median": 199.0,
            "p95_nearest_rank": 214,
            "min": 187,
            "max": 218
          },
          "actual_completion_tokens": {
            "n": 20,
            "median": 689.5,
            "p95_nearest_rank": 4403,
            "min": 243,
            "max": 4800
          },
          "effective_timeout_s": {
            "n": 20,
            "median": 1515.5912158475257,
            "p95_nearest_rank": 1515.592560135061,
            "min": 1515.588108799071,
            "max": 1515.5928013110533
          },
          "deadline_censored": 0,
          "requests_sha256": "396cd335d1f28aa7cee7f5c6eb54b09b304fac169d923012c57fdcfa5c71a589",
          "profile": {
            "max_tokens": 262144,
            "timeout_s": 5400,
            "chat_template_kwargs": {
              "thinking": true,
              "reasoning_effort": 100
            },
            "temperature": 1.0,
            "top_p": 0.95
          }
        },
        {
          "condition": "flash-fair262k-c20-short_science_qa",
          "concurrency": 20,
          "planned": 20,
          "started": 20,
          "complete": 20,
          "statuses": {
            "complete": 20
          },
          "automatic_pass": 20,
          "automatic_fail": 0,
          "rubric_only": 0,
          "unscored_incomplete": 0,
          "latency_s_complete_only": {
            "first_generated_s": {
              "n": 20,
              "median": 1.4815258929738775,
              "p95_nearest_rank": 1.4834380969405174,
              "min": 0.09093543700873852,
              "max": 1.4839434080058709
            },
            "first_final_s": {
              "n": 20,
              "median": 2.0148734245449305,
              "p95_nearest_rank": 2.4973933420842513,
              "min": 1.7942242129938677,
              "max": 2.7658712370321155
            },
            "completion_s": {
              "n": 20,
              "median": 2.8945703724748455,
              "p95_nearest_rank": 3.700675316969864,
              "min": 2.348468965035863,
              "max": 3.745721013052389
            }
          },
          "elapsed_s": 3.7470324110472575,
          "actual_prompt_tokens": {
            "n": 20,
            "median": 186.0,
            "p95_nearest_rank": 194,
            "min": 173,
            "max": 198
          },
          "actual_completion_tokens": {
            "n": 20,
            "median": 220.0,
            "p95_nearest_rank": 382,
            "min": 135,
            "max": 392
          },
          "effective_timeout_s": {
            "n": 20,
            "median": 1526.9851945135742,
            "p95_nearest_rank": 1526.9869858230231,
            "min": 1526.982994500082,
            "max": 1526.987243249081
          },
          "deadline_censored": 0,
          "requests_sha256": "f079ec7f45acbf50b9931c608cc7b4916fb0b96c15e9978ebfed1ccb13b48df0",
          "profile": {
            "max_tokens": 262144,
            "timeout_s": 5400,
            "chat_template_kwargs": {
              "thinking": true,
              "reasoning_effort": 25
            },
            "temperature": 1.0,
            "top_p": 0.95
          }
        },
        {
          "condition": "flash-fair262k-c5-long_document_grounded_qa",
          "concurrency": 5,
          "planned": 20,
          "started": 20,
          "complete": 20,
          "statuses": {
            "complete": 20
          },
          "automatic_pass": 20,
          "automatic_fail": 0,
          "rubric_only": 0,
          "unscored_incomplete": 0,
          "latency_s_complete_only": {
            "first_generated_s": {
              "n": 20,
              "median": 0.35877334902761504,
              "p95_nearest_rank": 1.4688742709113285,
              "min": 0.3532568559749052,
              "max": 1.8889481690712273
            },
            "first_final_s": {
              "n": 20,
              "median": 1.7224093364784494,
              "p95_nearest_rank": 3.275654525961727,
              "min": 0.824309010989964,
              "max": 3.4487887570867315
            },
            "completion_s": {
              "n": 20,
              "median": 3.226141944003757,
              "p95_nearest_rank": 5.2848796620965,
              "min": 1.8001638549612835,
              "max": 5.509042778983712
            }
          },
          "elapsed_s": 14.29237592604477,
          "actual_prompt_tokens": {
            "n": 20,
            "median": 17316.0,
            "p95_nearest_rank": 17316,
            "min": 17316,
            "max": 17316
          },
          "actual_completion_tokens": {
            "n": 20,
            "median": 305.0,
            "p95_nearest_rank": 348,
            "min": 253,
            "max": 402
          },
          "effective_timeout_s": {
            "n": 20,
            "median": 1615.9732127960306,
            "p95_nearest_rank": 1622.2321542120771,
            "min": 1609.740249946015,
            "max": 1622.232481016079
          },
          "deadline_censored": 0,
          "requests_sha256": "b55796fbdd87fd0bba20a097f52daed55b78b02d0eec175ee8d6419651077e07",
          "profile": {
            "max_tokens": 262144,
            "timeout_s": 5400,
            "chat_template_kwargs": {
              "thinking": true,
              "reasoning_effort": 25
            },
            "temperature": 1.0,
            "top_p": 0.95
          }
        },
        {
          "condition": "flash-fair262k-c5-scientific_reasoning",
          "concurrency": 5,
          "planned": 20,
          "started": 20,
          "complete": 20,
          "statuses": {
            "complete": 20
          },
          "automatic_pass": 12,
          "automatic_fail": 0,
          "rubric_only": 8,
          "unscored_incomplete": 0,
          "latency_s_complete_only": {
            "first_generated_s": {
              "n": 20,
              "median": 0.07277995650656521,
              "p95_nearest_rank": 0.14018646394833922,
              "min": 0.0706230680225417,
              "max": 0.14050113793928176
            },
            "first_final_s": {
              "n": 20,
              "median": 2.3664547154912725,
              "p95_nearest_rank": 17.650894510094076,
              "min": 0.5412435099715367,
              "max": 21.96358710306231
            },
            "completion_s": {
              "n": 20,
              "median": 3.722883523441851,
              "p95_nearest_rank": 19.326786105055362,
              "min": 1.5929365740157664,
              "max": 24.318846564972773
            }
          },
          "elapsed_s": 36.62509247590788,
          "actual_prompt_tokens": {
            "n": 20,
            "median": 199.0,
            "p95_nearest_rank": 214,
            "min": 187,
            "max": 218
          },
          "actual_completion_tokens": {
            "n": 20,
            "median": 642.5,
            "p95_nearest_rank": 3701,
            "min": 268,
            "max": 5097
          },
          "effective_timeout_s": {
            "n": 20,
            "median": 1603.693747549958,
            "p95_nearest_rank": 1607.9371480899863,
            "min": 1592.9164001198951,
            "max": 1607.937414365937
          },
          "deadline_censored": 0,
          "requests_sha256": "b9e433b2d3821fc5b019a1af36efcbd66bf8c97a3bff42c59f040f8458fe6e38",
          "profile": {
            "max_tokens": 262144,
            "timeout_s": 5400,
            "chat_template_kwargs": {
              "thinking": true,
              "reasoning_effort": 100
            },
            "temperature": 1.0,
            "top_p": 0.95
          }
        },
        {
          "condition": "flash-fair262k-c5-short_science_qa",
          "concurrency": 5,
          "planned": 20,
          "started": 20,
          "complete": 20,
          "statuses": {
            "complete": 20
          },
          "automatic_pass": 20,
          "automatic_fail": 0,
          "rubric_only": 0,
          "unscored_incomplete": 0,
          "latency_s_complete_only": {
            "first_generated_s": {
              "n": 20,
              "median": 0.07005949248559773,
              "p95_nearest_rank": 0.13787962193600833,
              "min": 0.06949802697636187,
              "max": 0.1385479710297659
            },
            "first_final_s": {
              "n": 20,
              "median": 0.6594311110093258,
              "p95_nearest_rank": 1.0045741440262645,
              "min": 0.3152666030218825,
              "max": 1.7766599990427494
            },
            "completion_s": {
              "n": 20,
              "median": 1.5170406795223244,
              "p95_nearest_rank": 2.3339831799967214,
              "min": 0.7997403229819611,
              "max": 2.5184605810791254
            }
          },
          "elapsed_s": 6.550185419968329,
          "actual_prompt_tokens": {
            "n": 20,
            "median": 186.0,
            "p95_nearest_rank": 194,
            "min": 173,
            "max": 198
          },
          "actual_completion_tokens": {
            "n": 20,
            "median": 244.0,
            "p95_nearest_rank": 380,
            "min": 135,
            "max": 394
          },
          "effective_timeout_s": {
            "n": 20,
            "median": 1626.6178974035429,
            "p95_nearest_rank": 1628.7849766770378,
            "min": 1623.4257236600388,
            "max": 1628.7852354530478
          },
          "deadline_censored": 0,
          "requests_sha256": "920eb778065aa4364cdfb5d416d969632fe23e7aa443521198bddaff7101ce8e",
          "profile": {
            "max_tokens": 262144,
            "timeout_s": 5400,
            "chat_template_kwargs": {
              "thinking": true,
              "reasoning_effort": 25
            },
            "temperature": 1.0,
            "top_p": 0.95
          }
        }
      ],
      "capacity": {
        "repo": "deepseek-ai/DeepSeek-V4.1-Flash-1017",
        "revision": "1bae0e1088993b9b82de08f59de5be5a34e42170",
        "configured_context_tokens": 1048576,
        "startup": {
          "ready": true,
          "seconds": 769.338428214076,
          "cumulative_seconds": 1659.4075004762271,
          "returncode": null
        },
        "benchmark_error": null,
        "planned_prompt_targets": [
          131072,
          262144,
          524288,
          786432
        ],
        "conditions": [
          {
            "condition": "flash-capacity-131072-c1",
            "concurrency": 1,
            "planned": 1,
            "returned": 1,
            "condition_closed": true,
            "not_returned": 0,
            "complete": 1,
            "numeric_citation_pass": 1,
            "statuses": {
              "complete": 1
            },
            "noncompletion_fraction_of_returned": 0.0,
            "completion_fraction_of_planned": 1.0,
            "successful_latency_s": {
              "first_generated_s": {
                "n": 1,
                "min": 2.4198986990377307,
                "median": 2.4198986990377307,
                "max": 2.4198986990377307
              },
              "first_final_s": {
                "n": 1,
                "min": 3.207754548988305,
                "median": 3.207754548988305,
                "max": 3.207754548988305
              },
              "completion_s": {
                "n": 1,
                "min": 3.9567935990635306,
                "median": 3.9567935990635306,
                "max": 3.9567935990635306
              }
            },
            "resources": {
              "sample_count": 0,
              "sampled_gpu_peak_mib_by_index": {},
              "minimum_host_memavailable_bytes": null,
              "qualification": "10-second samples during the condition; not an exact peak. Host available RAM is host-wide, not model allocation."
            },
            "rows": [
              {
                "request_id": "flash-capacity-131072-c1-0000",
                "case_id": "capacity-flash-131072-00",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130731,
                "actual_prompt_tokens": 130731,
                "actual_completion_tokens": 363,
                "actual_reasoning_tokens": 188,
                "actual_total_tokens": 131094,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 2.4198986990377307,
                "first_final_s": 3.207754548988305,
                "completion_s": 3.9567935990635306
              }
            ],
            "effective_timeout_s": {
              "n": 1,
              "median": 1799.6415401210543,
              "p95_nearest_rank": 1799.6415401210543,
              "min": 1799.6415401210543,
              "max": 1799.6415401210543
            },
            "deadline_censored": 0
          },
          {
            "condition": "flash-capacity-131072-c10",
            "concurrency": 10,
            "planned": 10,
            "returned": 10,
            "condition_closed": true,
            "not_returned": 0,
            "complete": 10,
            "numeric_citation_pass": 10,
            "statuses": {
              "complete": 10
            },
            "noncompletion_fraction_of_returned": 0.0,
            "completion_fraction_of_planned": 1.0,
            "successful_latency_s": {
              "first_generated_s": {
                "n": 10,
                "min": 2.4141667030053213,
                "median": 12.33240539644612,
                "max": 23.55696242803242
              },
              "first_final_s": {
                "n": 10,
                "min": 13.131062544998713,
                "median": 22.57844208041206,
                "max": 24.38030776300002
              },
              "completion_s": {
                "n": 10,
                "min": 23.796257931971923,
                "median": 24.625612611940596,
                "max": 25.353272174019367
              }
            },
            "resources": {
              "sample_count": 3,
              "sampled_gpu_peak_mib_by_index": {
                "0": 151082.0,
                "1": 150460.0,
                "2": 150460.0,
                "3": 150460.0,
                "4": 150460.0,
                "5": 150460.0,
                "6": 150460.0,
                "7": 150460.0
              },
              "minimum_host_memavailable_bytes": 3142374342656,
              "qualification": "10-second samples during the condition; not an exact peak. Host available RAM is host-wide, not model allocation."
            },
            "rows": [
              {
                "request_id": "flash-capacity-131072-c10-0001",
                "case_id": "capacity-flash-131072-01",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130731,
                "actual_prompt_tokens": 130731,
                "actual_completion_tokens": 298,
                "actual_reasoning_tokens": 123,
                "actual_total_tokens": 131029,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 4.6185862079728395,
                "first_final_s": 13.131062544998713,
                "completion_s": 23.796257931971923
              },
              {
                "request_id": "flash-capacity-131072-c10-0002",
                "case_id": "capacity-flash-131072-02",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130731,
                "actual_prompt_tokens": 130731,
                "actual_completion_tokens": 336,
                "actual_reasoning_tokens": 168,
                "actual_total_tokens": 131067,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 2.4141667030053213,
                "first_final_s": 14.008190001011826,
                "completion_s": 23.832196040078998
              },
              {
                "request_id": "flash-capacity-131072-c10-0000",
                "case_id": "capacity-flash-131072-00",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130731,
                "actual_prompt_tokens": 130731,
                "actual_completion_tokens": 356,
                "actual_reasoning_tokens": 156,
                "actual_total_tokens": 131087,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 6.828495269990526,
                "first_final_s": 17.627396508934908,
                "completion_s": 24.293936589965597
              },
              {
                "request_id": "flash-capacity-131072-c10-0006",
                "case_id": "capacity-flash-131072-06",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130731,
                "actual_prompt_tokens": 130731,
                "actual_completion_tokens": 317,
                "actual_reasoning_tokens": 150,
                "actual_total_tokens": 131048,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 11.230635521933436,
                "first_final_s": 21.60853290790692,
                "completion_s": 24.420791925978847
              },
              {
                "request_id": "flash-capacity-131072-c10-0007",
                "case_id": "capacity-flash-131072-07",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130731,
                "actual_prompt_tokens": 130731,
                "actual_completion_tokens": 385,
                "actual_reasoning_tokens": 173,
                "actual_total_tokens": 131116,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 9.022696704952978,
                "first_final_s": 20.9745494419476,
                "completion_s": 24.6138054439798
              },
              {
                "request_id": "flash-capacity-131072-c10-0009",
                "case_id": "capacity-flash-131072-09",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130731,
                "actual_prompt_tokens": 130731,
                "actual_completion_tokens": 326,
                "actual_reasoning_tokens": 129,
                "actual_total_tokens": 131057,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 13.434175270958804,
                "first_final_s": 23.5483512529172,
                "completion_s": 24.637419779901393
              },
              {
                "request_id": "flash-capacity-131072-c10-0008",
                "case_id": "capacity-flash-131072-08",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130731,
                "actual_prompt_tokens": 130731,
                "actual_completion_tokens": 264,
                "actual_reasoning_tokens": 112,
                "actual_total_tokens": 130995,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 20.07075705297757,
                "first_final_s": 24.007468944997527,
                "completion_s": 24.803452719002962
              },
              {
                "request_id": "flash-capacity-131072-c10-0004",
                "case_id": "capacity-flash-131072-04",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130731,
                "actual_prompt_tokens": 130731,
                "actual_completion_tokens": 391,
                "actual_reasoning_tokens": 187,
                "actual_total_tokens": 131122,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 15.652596820960753,
                "first_final_s": 24.07337687292602,
                "completion_s": 25.115711171994917
              },
              {
                "request_id": "flash-capacity-131072-c10-0005",
                "case_id": "capacity-flash-131072-05",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130731,
                "actual_prompt_tokens": 130731,
                "actual_completion_tokens": 359,
                "actual_reasoning_tokens": 196,
                "actual_total_tokens": 131090,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 17.861987472046167,
                "first_final_s": 24.29518023901619,
                "completion_s": 25.114903716021217
              },
              {
                "request_id": "flash-capacity-131072-c10-0003",
                "case_id": "capacity-flash-131072-03",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130731,
                "actual_prompt_tokens": 130731,
                "actual_completion_tokens": 350,
                "actual_reasoning_tokens": 147,
                "actual_total_tokens": 131081,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 23.55696242803242,
                "first_final_s": 24.38030776300002,
                "completion_s": 25.353272174019367
              }
            ],
            "effective_timeout_s": {
              "n": 10,
              "median": 1725.901182607573,
              "p95_nearest_rank": 1725.9082933340687,
              "min": 1725.894195870147,
              "max": 1725.9082933340687
            },
            "deadline_censored": 0
          },
          {
            "condition": "flash-capacity-131072-c20",
            "concurrency": 20,
            "planned": 20,
            "returned": 20,
            "condition_closed": true,
            "not_returned": 0,
            "complete": 20,
            "numeric_citation_pass": 20,
            "statuses": {
              "complete": 20
            },
            "noncompletion_fraction_of_returned": 0.0,
            "completion_fraction_of_planned": 1.0,
            "successful_latency_s": {
              "first_generated_s": {
                "n": 20,
                "min": 2.427316203014925,
                "median": 23.428041853010654,
                "max": 44.470294194994494
              },
              "first_final_s": {
                "n": 20,
                "min": 11.002297321916558,
                "median": 34.83625016850419,
                "max": 45.46596932504326
              },
              "completion_s": {
                "n": 20,
                "min": 23.651014070026577,
                "median": 44.656639870489016,
                "max": 46.40721051697619
              }
            },
            "resources": {
              "sample_count": 4,
              "sampled_gpu_peak_mib_by_index": {
                "0": 151082.0,
                "1": 150460.0,
                "2": 150460.0,
                "3": 150460.0,
                "4": 150460.0,
                "5": 150460.0,
                "6": 150460.0,
                "7": 150460.0
              },
              "minimum_host_memavailable_bytes": 3141663596544,
              "qualification": "10-second samples during the condition; not an exact peak. Host available RAM is host-wide, not model allocation."
            },
            "rows": [
              {
                "request_id": "flash-capacity-131072-c20-0006",
                "case_id": "capacity-flash-131072-06",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130731,
                "actual_prompt_tokens": 130731,
                "actual_completion_tokens": 309,
                "actual_reasoning_tokens": 124,
                "actual_total_tokens": 131040,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 2.427316203014925,
                "first_final_s": 11.002297321916558,
                "completion_s": 23.651014070026577
              },
              {
                "request_id": "flash-capacity-131072-c20-0008",
                "case_id": "capacity-flash-131072-08",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130731,
                "actual_prompt_tokens": 130731,
                "actual_completion_tokens": 284,
                "actual_reasoning_tokens": 133,
                "actual_total_tokens": 131015,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 4.626475075026974,
                "first_final_s": 13.812921578064561,
                "completion_s": 24.153677835012786
              },
              {
                "request_id": "flash-capacity-131072-c20-0003",
                "case_id": "capacity-flash-131072-03",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130731,
                "actual_prompt_tokens": 130731,
                "actual_completion_tokens": 314,
                "actual_reasoning_tokens": 138,
                "actual_total_tokens": 131045,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 9.041916083078831,
                "first_final_s": 18.589920182013884,
                "completion_s": 30.65642855607439
              },
              {
                "request_id": "flash-capacity-131072-c20-0001",
                "case_id": "capacity-flash-131072-01",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130731,
                "actual_prompt_tokens": 130731,
                "actual_completion_tokens": 347,
                "actual_reasoning_tokens": 155,
                "actual_total_tokens": 131078,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 6.837293572025374,
                "first_final_s": 17.578999817022122,
                "completion_s": 30.732429784024134
              },
              {
                "request_id": "flash-capacity-131072-c20-0010",
                "case_id": "capacity-flash-131072-10",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130731,
                "actual_prompt_tokens": 130731,
                "actual_completion_tokens": 343,
                "actual_reasoning_tokens": 194,
                "actual_total_tokens": 131074,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 11.249106969917193,
                "first_final_s": 24.70623883290682,
                "completion_s": 34.8598423829535
              },
              {
                "request_id": "flash-capacity-131072-c20-0015",
                "case_id": "capacity-flash-131072-15",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130731,
                "actual_prompt_tokens": 130731,
                "actual_completion_tokens": 265,
                "actual_reasoning_tokens": 113,
                "actual_total_tokens": 130996,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 17.873924268991686,
                "first_final_s": 25.709898659959435,
                "completion_s": 36.10152942396235
              },
              {
                "request_id": "flash-capacity-131072-c20-0017",
                "case_id": "capacity-flash-131072-17",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130731,
                "actual_prompt_tokens": 130731,
                "actual_completion_tokens": 345,
                "actual_reasoning_tokens": 156,
                "actual_total_tokens": 131076,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 13.435717812972143,
                "first_final_s": 24.285287400940433,
                "completion_s": 37.207668495946564
              },
              {
                "request_id": "flash-capacity-131072-c20-0014",
                "case_id": "capacity-flash-131072-14",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130731,
                "actual_prompt_tokens": 130731,
                "actual_completion_tokens": 378,
                "actual_reasoning_tokens": 175,
                "actual_total_tokens": 131109,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 15.657530326046981,
                "first_final_s": 27.784731394029222,
                "completion_s": 41.717706948984414
              },
              {
                "request_id": "flash-capacity-131072-c20-0000",
                "case_id": "capacity-flash-131072-00",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130731,
                "actual_prompt_tokens": 130731,
                "actual_completion_tokens": 356,
                "actual_reasoning_tokens": 173,
                "actual_total_tokens": 131087,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 20.115845186053775,
                "first_final_s": 32.0972965140827,
                "completion_s": 44.496063144062646
              },
              {
                "request_id": "flash-capacity-131072-c20-0002",
                "case_id": "capacity-flash-131072-02",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130731,
                "actual_prompt_tokens": 130731,
                "actual_completion_tokens": 270,
                "actual_reasoning_tokens": 123,
                "actual_total_tokens": 131001,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 26.763328324072063,
                "first_final_s": 35.30788419197779,
                "completion_s": 44.55359922000207
              },
              {
                "request_id": "flash-capacity-131072-c20-0011",
                "case_id": "capacity-flash-131072-11",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130731,
                "actual_prompt_tokens": 130731,
                "actual_completion_tokens": 369,
                "actual_reasoning_tokens": 174,
                "actual_total_tokens": 131100,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 22.32058734400198,
                "first_final_s": 34.364616145030595,
                "completion_s": 44.75968052097596
              },
              {
                "request_id": "flash-capacity-131072-c20-0013",
                "case_id": "capacity-flash-131072-13",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130731,
                "actual_prompt_tokens": 130731,
                "actual_completion_tokens": 348,
                "actual_reasoning_tokens": 166,
                "actual_total_tokens": 131079,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 24.53549636201933,
                "first_final_s": 36.037205580971204,
                "completion_s": 44.82539958995767
              },
              {
                "request_id": "flash-capacity-131072-c20-0005",
                "case_id": "capacity-flash-131072-05",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130731,
                "actual_prompt_tokens": 130731,
                "actual_completion_tokens": 299,
                "actual_reasoning_tokens": 126,
                "actual_total_tokens": 131030,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 28.972552597988397,
                "first_final_s": 37.74138526094612,
                "completion_s": 44.92026529205032
              },
              {
                "request_id": "flash-capacity-131072-c20-0018",
                "case_id": "capacity-flash-131072-18",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130731,
                "actual_prompt_tokens": 130731,
                "actual_completion_tokens": 304,
                "actual_reasoning_tokens": 132,
                "actual_total_tokens": 131035,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 35.598906269995496,
                "first_final_s": 44.47637458401732,
                "completion_s": 45.45317238999996
              },
              {
                "request_id": "flash-capacity-131072-c20-0019",
                "case_id": "capacity-flash-131072-19",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130731,
                "actual_prompt_tokens": 130731,
                "actual_completion_tokens": 364,
                "actual_reasoning_tokens": 164,
                "actual_total_tokens": 131095,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 33.37893940601498,
                "first_final_s": 44.473557171062566,
                "completion_s": 45.601942809065804
              },
              {
                "request_id": "flash-capacity-131072-c20-0012",
                "case_id": "capacity-flash-131072-12",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130731,
                "actual_prompt_tokens": 130731,
                "actual_completion_tokens": 397,
                "actual_reasoning_tokens": 195,
                "actual_total_tokens": 131128,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 31.17819869099185,
                "first_final_s": 44.48628390801605,
                "completion_s": 45.62008523498662
              },
              {
                "request_id": "flash-capacity-131072-c20-0007",
                "case_id": "capacity-flash-131072-07",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130731,
                "actual_prompt_tokens": 130731,
                "actual_completion_tokens": 362,
                "actual_reasoning_tokens": 179,
                "actual_total_tokens": 131093,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 37.833637256058864,
                "first_final_s": 44.97024639497977,
                "completion_s": 45.92186723300256
              },
              {
                "request_id": "flash-capacity-131072-c20-0004",
                "case_id": "capacity-flash-131072-04",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130731,
                "actual_prompt_tokens": 130731,
                "actual_completion_tokens": 364,
                "actual_reasoning_tokens": 173,
                "actual_total_tokens": 131095,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 40.050501700956374,
                "first_final_s": 45.11762147699483,
                "completion_s": 46.09217793098651
              },
              {
                "request_id": "flash-capacity-131072-c20-0009",
                "case_id": "capacity-flash-131072-09",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130731,
                "actual_prompt_tokens": 130731,
                "actual_completion_tokens": 370,
                "actual_reasoning_tokens": 182,
                "actual_total_tokens": 131101,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 42.261608953005634,
                "first_final_s": 45.334643734036945,
                "completion_s": 46.25891058007255
              },
              {
                "request_id": "flash-capacity-131072-c20-0016",
                "case_id": "capacity-flash-131072-16",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130731,
                "actual_prompt_tokens": 130731,
                "actual_completion_tokens": 375,
                "actual_reasoning_tokens": 176,
                "actual_total_tokens": 131106,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 44.470294194994494,
                "first_final_s": 45.46596932504326,
                "completion_s": 46.40721051697619
              }
            ],
            "effective_timeout_s": {
              "n": 20,
              "median": 1700.5316147204721,
              "p95_nearest_rank": 1700.5446064529242,
              "min": 1700.5151483359514,
              "max": 1700.546337403939
            },
            "deadline_censored": 0
          },
          {
            "condition": "flash-capacity-131072-c5",
            "concurrency": 5,
            "planned": 5,
            "returned": 5,
            "condition_closed": true,
            "not_returned": 0,
            "complete": 5,
            "numeric_citation_pass": 5,
            "statuses": {
              "complete": 5
            },
            "noncompletion_fraction_of_returned": 0.0,
            "completion_fraction_of_planned": 1.0,
            "successful_latency_s": {
              "first_generated_s": {
                "n": 5,
                "min": 2.4292777410009876,
                "median": 6.8528362029464915,
                "max": 11.265536888036877
              },
              "first_final_s": {
                "n": 5,
                "min": 11.267904678941704,
                "median": 11.781498839962296,
                "max": 12.003433704958297
              },
              "completion_s": {
                "n": 5,
                "min": 12.182509226957336,
                "median": 12.702863384969532,
                "max": 12.95221413008403
              }
            },
            "resources": {
              "sample_count": 1,
              "sampled_gpu_peak_mib_by_index": {
                "0": 151082.0,
                "1": 150460.0,
                "2": 150460.0,
                "3": 150460.0,
                "4": 150460.0,
                "5": 150460.0,
                "6": 150460.0,
                "7": 150460.0
              },
              "minimum_host_memavailable_bytes": 3141360611328,
              "qualification": "10-second samples during the condition; not an exact peak. Host available RAM is host-wide, not model allocation."
            },
            "rows": [
              {
                "request_id": "flash-capacity-131072-c5-0003",
                "case_id": "capacity-flash-131072-03",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130731,
                "actual_prompt_tokens": 130731,
                "actual_completion_tokens": 303,
                "actual_reasoning_tokens": 128,
                "actual_total_tokens": 131034,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 2.4292777410009876,
                "first_final_s": 11.267904678941704,
                "completion_s": 12.182509226957336
              },
              {
                "request_id": "flash-capacity-131072-c5-0000",
                "case_id": "capacity-flash-131072-00",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130731,
                "actual_prompt_tokens": 130731,
                "actual_completion_tokens": 331,
                "actual_reasoning_tokens": 166,
                "actual_total_tokens": 131062,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 4.645337260095403,
                "first_final_s": 11.63401961803902,
                "completion_s": 12.4779376500519
              },
              {
                "request_id": "flash-capacity-131072-c5-0002",
                "case_id": "capacity-flash-131072-02",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130731,
                "actual_prompt_tokens": 130731,
                "actual_completion_tokens": 346,
                "actual_reasoning_tokens": 162,
                "actual_total_tokens": 131077,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 6.8528362029464915,
                "first_final_s": 11.781498839962296,
                "completion_s": 12.702863384969532
              },
              {
                "request_id": "flash-capacity-131072-c5-0001",
                "case_id": "capacity-flash-131072-01",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130731,
                "actual_prompt_tokens": 130731,
                "actual_completion_tokens": 364,
                "actual_reasoning_tokens": 171,
                "actual_total_tokens": 131095,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 9.063928605988622,
                "first_final_s": 12.003433704958297,
                "completion_s": 12.932771325926296
              },
              {
                "request_id": "flash-capacity-131072-c5-0004",
                "case_id": "capacity-flash-131072-04",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130731,
                "actual_prompt_tokens": 130731,
                "actual_completion_tokens": 338,
                "actual_reasoning_tokens": 139,
                "actual_total_tokens": 131069,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 11.265536888036877,
                "first_final_s": 11.99762645107694,
                "completion_s": 12.95221413008403
              }
            ],
            "effective_timeout_s": {
              "n": 5,
              "median": 1738.8673406291055,
              "p95_nearest_rank": 1738.870732702082,
              "min": 1738.863029591157,
              "max": 1738.870732702082
            },
            "deadline_censored": 0
          },
          {
            "condition": "flash-capacity-262144-c1",
            "concurrency": 1,
            "planned": 1,
            "returned": 1,
            "condition_closed": true,
            "not_returned": 0,
            "complete": 1,
            "numeric_citation_pass": 1,
            "statuses": {
              "complete": 1
            },
            "noncompletion_fraction_of_returned": 0.0,
            "completion_fraction_of_planned": 1.0,
            "successful_latency_s": {
              "first_generated_s": {
                "n": 1,
                "min": 5.248314737924375,
                "median": 5.248314737924375,
                "max": 5.248314737924375
              },
              "first_final_s": {
                "n": 1,
                "min": 5.850156461936422,
                "median": 5.850156461936422,
                "max": 5.850156461936422
              },
              "completion_s": {
                "n": 1,
                "min": 6.515659660915844,
                "median": 6.515659660915844,
                "max": 6.515659660915844
              }
            },
            "resources": {
              "sample_count": 1,
              "sampled_gpu_peak_mib_by_index": {
                "0": 151074.0,
                "1": 150452.0,
                "2": 150452.0,
                "3": 150452.0,
                "4": 150452.0,
                "5": 150452.0,
                "6": 150452.0,
                "7": 150452.0
              },
              "minimum_host_memavailable_bytes": 3140442714112,
              "qualification": "10-second samples during the condition; not an exact peak. Host available RAM is host-wide, not model allocation."
            },
            "rows": [
              {
                "request_id": "flash-capacity-262144-c1-0000",
                "case_id": "capacity-flash-262144-00",
                "target_prompt_tokens": 262144,
                "tokenizer_prompt_tokens": 261558,
                "actual_prompt_tokens": 261558,
                "actual_completion_tokens": 304,
                "actual_reasoning_tokens": 149,
                "actual_total_tokens": 261862,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 5.248314737924375,
                "first_final_s": 5.850156461936422,
                "completion_s": 6.515659660915844
              }
            ],
            "effective_timeout_s": {
              "n": 1,
              "median": 1794.9102802370908,
              "p95_nearest_rank": 1794.9102802370908,
              "min": 1794.9102802370908,
              "max": 1794.9102802370908
            },
            "deadline_censored": 0
          },
          {
            "condition": "flash-capacity-524288-c1",
            "concurrency": 1,
            "planned": 1,
            "returned": 1,
            "condition_closed": true,
            "not_returned": 0,
            "complete": 1,
            "numeric_citation_pass": 1,
            "statuses": {
              "complete": 1
            },
            "noncompletion_fraction_of_returned": 0.0,
            "completion_fraction_of_planned": 1.0,
            "successful_latency_s": {
              "first_generated_s": {
                "n": 1,
                "min": 12.417156481998973,
                "median": 12.417156481998973,
                "max": 12.417156481998973
              },
              "first_final_s": {
                "n": 1,
                "min": 13.045501222950406,
                "median": 13.045501222950406,
                "max": 13.045501222950406
              },
              "completion_s": {
                "n": 1,
                "min": 13.799586507957429,
                "median": 13.799586507957429,
                "max": 13.799586507957429
              }
            },
            "resources": {
              "sample_count": 2,
              "sampled_gpu_peak_mib_by_index": {
                "0": 151082.0,
                "1": 150460.0,
                "2": 150460.0,
                "3": 150460.0,
                "4": 150460.0,
                "5": 150460.0,
                "6": 150460.0,
                "7": 150460.0
              },
              "minimum_host_memavailable_bytes": 3141278126080,
              "qualification": "10-second samples during the condition; not an exact peak. Host available RAM is host-wide, not model allocation."
            },
            "rows": [
              {
                "request_id": "flash-capacity-524288-c1-0000",
                "case_id": "capacity-flash-524288-00",
                "target_prompt_tokens": 524288,
                "tokenizer_prompt_tokens": 523717,
                "actual_prompt_tokens": 523717,
                "actual_completion_tokens": 339,
                "actual_reasoning_tokens": 165,
                "actual_total_tokens": 524056,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 12.417156481998973,
                "first_final_s": 13.045501222950406,
                "completion_s": 13.799586507957429
              }
            ],
            "effective_timeout_s": {
              "n": 1,
              "median": 1786.5738587169908,
              "p95_nearest_rank": 1786.5738587169908,
              "min": 1786.5738587169908,
              "max": 1786.5738587169908
            },
            "deadline_censored": 0
          },
          {
            "condition": "flash-capacity-786432-c1",
            "concurrency": 1,
            "planned": 1,
            "returned": 1,
            "condition_closed": true,
            "not_returned": 0,
            "complete": 1,
            "numeric_citation_pass": 1,
            "statuses": {
              "complete": 1
            },
            "noncompletion_fraction_of_returned": 0.0,
            "completion_fraction_of_planned": 1.0,
            "successful_latency_s": {
              "first_generated_s": {
                "n": 1,
                "min": 23.44271283701528,
                "median": 23.44271283701528,
                "max": 23.44271283701528
              },
              "first_final_s": {
                "n": 1,
                "min": 23.738149481010623,
                "median": 23.738149481010623,
                "max": 23.738149481010623
              },
              "completion_s": {
                "n": 1,
                "min": 24.47663565701805,
                "median": 24.47663565701805,
                "max": 24.47663565701805
              }
            },
            "resources": {
              "sample_count": 2,
              "sampled_gpu_peak_mib_by_index": {
                "0": 151082.0,
                "1": 150460.0,
                "2": 150460.0,
                "3": 150460.0,
                "4": 150460.0,
                "5": 150460.0,
                "6": 150460.0,
                "7": 150460.0
              },
              "minimum_host_memavailable_bytes": 3142534287360,
              "qualification": "10-second samples during the condition; not an exact peak. Host available RAM is host-wide, not model allocation."
            },
            "rows": [
              {
                "request_id": "flash-capacity-786432-c1-0000",
                "case_id": "capacity-flash-786432-00",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785876,
                "actual_prompt_tokens": 785876,
                "actual_completion_tokens": 268,
                "actual_reasoning_tokens": 99,
                "actual_total_tokens": 786144,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 23.44271283701528,
                "first_final_s": 23.738149481010623,
                "completion_s": 24.47663565701805
              }
            ],
            "effective_timeout_s": {
              "n": 1,
              "median": 1769.956793228048,
              "p95_nearest_rank": 1769.956793228048,
              "min": 1769.956793228048,
              "max": 1769.956793228048
            },
            "deadline_censored": 0
          },
          {
            "condition": "flash-capacity-786432-c10",
            "concurrency": 10,
            "planned": 10,
            "returned": 10,
            "condition_closed": true,
            "not_returned": 0,
            "complete": 10,
            "numeric_citation_pass": 10,
            "statuses": {
              "complete": 10
            },
            "noncompletion_fraction_of_returned": 0.0,
            "completion_fraction_of_planned": 1.0,
            "successful_latency_s": {
              "first_generated_s": {
                "n": 10,
                "min": 22.173694221070036,
                "median": 114.68718561000424,
                "max": 207.15134829701856
              },
              "first_final_s": {
                "n": 10,
                "min": 38.34419626102317,
                "median": 131.31023216748144,
                "max": 207.82543805008754
              },
              "completion_s": {
                "n": 10,
                "min": 52.34191706799902,
                "median": 148.23184268100886,
                "max": 208.44921233598143
              }
            },
            "resources": {
              "sample_count": 20,
              "sampled_gpu_peak_mib_by_index": {
                "0": 151082.0,
                "1": 150460.0,
                "2": 150460.0,
                "3": 150460.0,
                "4": 150460.0,
                "5": 150460.0,
                "6": 150460.0,
                "7": 150460.0
              },
              "minimum_host_memavailable_bytes": 3143491014656,
              "qualification": "10-second samples during the condition; not an exact peak. Host available RAM is host-wide, not model allocation."
            },
            "rows": [
              {
                "request_id": "flash-capacity-786432-c10-0002",
                "case_id": "capacity-flash-786432-02",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785876,
                "actual_prompt_tokens": 785876,
                "actual_completion_tokens": 306,
                "actual_reasoning_tokens": 164,
                "actual_total_tokens": 786182,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 22.173694221070036,
                "first_final_s": 38.34419626102317,
                "completion_s": 52.34191706799902
              },
              {
                "request_id": "flash-capacity-786432-c10-0005",
                "case_id": "capacity-flash-786432-05",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785876,
                "actual_prompt_tokens": 785876,
                "actual_completion_tokens": 361,
                "actual_reasoning_tokens": 212,
                "actual_total_tokens": 786237,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 42.7267114239512,
                "first_final_s": 64.55914792593103,
                "completion_s": 79.91624891199172
              },
              {
                "request_id": "flash-capacity-786432-c10-0003",
                "case_id": "capacity-flash-786432-03",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785876,
                "actual_prompt_tokens": 785876,
                "actual_completion_tokens": 274,
                "actual_reasoning_tokens": 114,
                "actual_total_tokens": 786150,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 63.30699774099048,
                "first_final_s": 73.114146491047,
                "completion_s": 90.10646194801666
              },
              {
                "request_id": "flash-capacity-786432-c10-0004",
                "case_id": "capacity-flash-786432-04",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785876,
                "actual_prompt_tokens": 785876,
                "actual_completion_tokens": 332,
                "actual_reasoning_tokens": 155,
                "actual_total_tokens": 786208,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 83.8795218860032,
                "first_final_s": 98.77628935792018,
                "completion_s": 117.09714027994778
              },
              {
                "request_id": "flash-capacity-786432-c10-0006",
                "case_id": "capacity-flash-786432-06",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785876,
                "actual_prompt_tokens": 785876,
                "actual_completion_tokens": 330,
                "actual_reasoning_tokens": 162,
                "actual_total_tokens": 786206,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 104.42443572008051,
                "first_final_s": 120.30060323304497,
                "completion_s": 137.38378849707078
              },
              {
                "request_id": "flash-capacity-786432-c10-0008",
                "case_id": "capacity-flash-786432-08",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785876,
                "actual_prompt_tokens": 785876,
                "actual_completion_tokens": 339,
                "actual_reasoning_tokens": 172,
                "actual_total_tokens": 786215,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 124.94993549992796,
                "first_final_s": 142.31986110191792,
                "completion_s": 159.07989686494693
              },
              {
                "request_id": "flash-capacity-786432-c10-0000",
                "case_id": "capacity-flash-786432-00",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785876,
                "actual_prompt_tokens": 785876,
                "actual_completion_tokens": 366,
                "actual_reasoning_tokens": 197,
                "actual_total_tokens": 786242,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 145.59154077304993,
                "first_final_s": 166.40823654306587,
                "completion_s": 183.5491513670422
              },
              {
                "request_id": "flash-capacity-786432-c10-0001",
                "case_id": "capacity-flash-786432-01",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785876,
                "actual_prompt_tokens": 785876,
                "actual_completion_tokens": 394,
                "actual_reasoning_tokens": 197,
                "actual_total_tokens": 786270,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 166.1475912469905,
                "first_final_s": 186.96291518001817,
                "completion_s": 207.22059867798816
              },
              {
                "request_id": "flash-capacity-786432-c10-0009",
                "case_id": "capacity-flash-786432-09",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785876,
                "actual_prompt_tokens": 785876,
                "actual_completion_tokens": 324,
                "actual_reasoning_tokens": 167,
                "actual_total_tokens": 786200,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 186.630104921991,
                "first_final_s": 203.25267854402773,
                "completion_s": 207.6225844270084
              },
              {
                "request_id": "flash-capacity-786432-c10-0007",
                "case_id": "capacity-flash-786432-07",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785876,
                "actual_prompt_tokens": 785876,
                "actual_completion_tokens": 316,
                "actual_reasoning_tokens": 173,
                "actual_total_tokens": 786192,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 207.15134829701856,
                "first_final_s": 207.82543805008754,
                "completion_s": 208.44921233598143
              }
            ],
            "effective_timeout_s": {
              "n": 10,
              "median": 1496.55083217856,
              "p95_nearest_rank": 1496.594402949093,
              "min": 1496.5001309370855,
              "max": 1496.594402949093
            },
            "deadline_censored": 0
          },
          {
            "condition": "flash-capacity-786432-c20",
            "concurrency": 20,
            "planned": 20,
            "returned": 20,
            "condition_closed": true,
            "not_returned": 0,
            "complete": 20,
            "numeric_citation_pass": 20,
            "statuses": {
              "complete": 20
            },
            "noncompletion_fraction_of_returned": 0.0,
            "completion_fraction_of_planned": 1.0,
            "successful_latency_s": {
              "first_generated_s": {
                "n": 20,
                "min": 22.125897224992514,
                "median": 217.6345697294455,
                "max": 412.8465921600582
              },
              "first_final_s": {
                "n": 20,
                "min": 33.75579446996562,
                "median": 230.86253121646587,
                "max": 413.4505691070808
              },
              "completion_s": {
                "n": 20,
                "min": 49.50283551006578,
                "median": 249.0319755434757,
                "max": 414.3025761710014
              }
            },
            "resources": {
              "sample_count": 41,
              "sampled_gpu_peak_mib_by_index": {
                "0": 151082.0,
                "1": 150460.0,
                "2": 150460.0,
                "3": 150460.0,
                "4": 150460.0,
                "5": 150460.0,
                "6": 150460.0,
                "7": 150460.0
              },
              "minimum_host_memavailable_bytes": 3146468683776,
              "qualification": "10-second samples during the condition; not an exact peak. Host available RAM is host-wide, not model allocation."
            },
            "rows": [
              {
                "request_id": "flash-capacity-786432-c20-0012",
                "case_id": "capacity-flash-786432-12",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785876,
                "actual_prompt_tokens": 785876,
                "actual_completion_tokens": 280,
                "actual_reasoning_tokens": 130,
                "actual_total_tokens": 786156,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 22.125897224992514,
                "first_final_s": 33.75579446996562,
                "completion_s": 49.50283551006578
              },
              {
                "request_id": "flash-capacity-786432-c20-0001",
                "case_id": "capacity-flash-786432-01",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785876,
                "actual_prompt_tokens": 785876,
                "actual_completion_tokens": 299,
                "actual_reasoning_tokens": 136,
                "actual_total_tokens": 786175,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 42.81042656500358,
                "first_final_s": 55.20433701691218,
                "completion_s": 72.20841392793227
              },
              {
                "request_id": "flash-capacity-786432-c20-0002",
                "case_id": "capacity-flash-786432-02",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785876,
                "actual_prompt_tokens": 785876,
                "actual_completion_tokens": 367,
                "actual_reasoning_tokens": 197,
                "actual_total_tokens": 786243,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 63.37218916800339,
                "first_final_s": 84.2044735099189,
                "completion_s": 101.50059054000303
              },
              {
                "request_id": "flash-capacity-786432-c20-0015",
                "case_id": "capacity-flash-786432-15",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785876,
                "actual_prompt_tokens": 785876,
                "actual_completion_tokens": 421,
                "actual_reasoning_tokens": 254,
                "actual_total_tokens": 786297,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 83.82637958100531,
                "first_final_s": 108.99122769909445,
                "completion_s": 127.36454750201665
              },
              {
                "request_id": "flash-capacity-786432-c20-0000",
                "case_id": "capacity-flash-786432-00",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785876,
                "actual_prompt_tokens": 785876,
                "actual_completion_tokens": 311,
                "actual_reasoning_tokens": 121,
                "actual_total_tokens": 786187,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 104.55785190104507,
                "first_final_s": 115.15363762900233,
                "completion_s": 135.27714006602764
              },
              {
                "request_id": "flash-capacity-786432-c20-0017",
                "case_id": "capacity-flash-786432-17",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785876,
                "actual_prompt_tokens": 785876,
                "actual_completion_tokens": 327,
                "actual_reasoning_tokens": 172,
                "actual_total_tokens": 786203,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 124.93798421195243,
                "first_final_s": 142.31879089295398,
                "completion_s": 157.55187651398592
              },
              {
                "request_id": "flash-capacity-786432-c20-0013",
                "case_id": "capacity-flash-786432-13",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785876,
                "actual_prompt_tokens": 785876,
                "actual_completion_tokens": 308,
                "actual_reasoning_tokens": 167,
                "actual_total_tokens": 786184,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 145.5591290970333,
                "first_final_s": 162.19941375602502,
                "completion_s": 175.9769844179973
              },
              {
                "request_id": "flash-capacity-786432-c20-0016",
                "case_id": "capacity-flash-786432-16",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785876,
                "actual_prompt_tokens": 785876,
                "actual_completion_tokens": 337,
                "actual_reasoning_tokens": 161,
                "actual_total_tokens": 786213,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 166.10394395899493,
                "first_final_s": 181.86040645604953,
                "completion_s": 199.98221280798316
              },
              {
                "request_id": "flash-capacity-786432-c20-0008",
                "case_id": "capacity-flash-786432-08",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785876,
                "actual_prompt_tokens": 785876,
                "actual_completion_tokens": 327,
                "actual_reasoning_tokens": 175,
                "actual_total_tokens": 786203,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 186.76356492808554,
                "first_final_s": 204.61658441007603,
                "completion_s": 219.37041847407818
              },
              {
                "request_id": "flash-capacity-786432-c20-0007",
                "case_id": "capacity-flash-786432-07",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785876,
                "actual_prompt_tokens": 785876,
                "actual_completion_tokens": 292,
                "actual_reasoning_tokens": 123,
                "actual_total_tokens": 786168,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 207.33947446895763,
                "first_final_s": 218.16767500294372,
                "completion_s": 235.97061552200466
              },
              {
                "request_id": "flash-capacity-786432-c20-0004",
                "case_id": "capacity-flash-786432-04",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785876,
                "actual_prompt_tokens": 785876,
                "actual_completion_tokens": 339,
                "actual_reasoning_tokens": 160,
                "actual_total_tokens": 786215,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 227.92966498993337,
                "first_final_s": 243.557387429988,
                "completion_s": 262.09333556494676
              },
              {
                "request_id": "flash-capacity-786432-c20-0005",
                "case_id": "capacity-flash-786432-05",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785876,
                "actual_prompt_tokens": 785876,
                "actual_completion_tokens": 284,
                "actual_reasoning_tokens": 138,
                "actual_total_tokens": 786160,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 248.48833326692693,
                "first_final_s": 261.1625915499171,
                "completion_s": 276.2858094769763
              },
              {
                "request_id": "flash-capacity-786432-c20-0011",
                "case_id": "capacity-flash-786432-11",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785876,
                "actual_prompt_tokens": 785876,
                "actual_completion_tokens": 326,
                "actual_reasoning_tokens": 166,
                "actual_total_tokens": 786202,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 268.99621699296404,
                "first_final_s": 285.4922181550646,
                "completion_s": 301.5057271940168
              },
              {
                "request_id": "flash-capacity-786432-c20-0014",
                "case_id": "capacity-flash-786432-14",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785876,
                "actual_prompt_tokens": 785876,
                "actual_completion_tokens": 364,
                "actual_reasoning_tokens": 177,
                "actual_total_tokens": 786240,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 289.52621670998633,
                "first_final_s": 307.7405392290093,
                "completion_s": 327.2558795019286
              },
              {
                "request_id": "flash-capacity-786432-c20-0006",
                "case_id": "capacity-flash-786432-06",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785876,
                "actual_prompt_tokens": 785876,
                "actual_completion_tokens": 353,
                "actual_reasoning_tokens": 202,
                "actual_total_tokens": 786229,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 310.1978691840777,
                "first_final_s": 331.35990213404875,
                "completion_s": 346.27651298907585
              },
              {
                "request_id": "flash-capacity-786432-c20-0019",
                "case_id": "capacity-flash-786432-19",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785876,
                "actual_prompt_tokens": 785876,
                "actual_completion_tokens": 313,
                "actual_reasoning_tokens": 169,
                "actual_total_tokens": 786189,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 330.53336446301546,
                "first_final_s": 347.40701913891826,
                "completion_s": 361.46282622695435
              },
              {
                "request_id": "flash-capacity-786432-c20-0010",
                "case_id": "capacity-flash-786432-10",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785876,
                "actual_prompt_tokens": 785876,
                "actual_completion_tokens": 300,
                "actual_reasoning_tokens": 155,
                "actual_total_tokens": 786176,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 351.22976476105396,
                "first_final_s": 366.1112973029958,
                "completion_s": 380.7403142930707
              },
              {
                "request_id": "flash-capacity-786432-c20-0003",
                "case_id": "capacity-flash-786432-03",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785876,
                "actual_prompt_tokens": 785876,
                "actual_completion_tokens": 306,
                "actual_reasoning_tokens": 121,
                "actual_total_tokens": 786182,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 371.87311490706634,
                "first_final_s": 382.46851888496894,
                "completion_s": 402.05545806803275
              },
              {
                "request_id": "flash-capacity-786432-c20-0009",
                "case_id": "capacity-flash-786432-09",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785876,
                "actual_prompt_tokens": 785876,
                "actual_completion_tokens": 366,
                "actual_reasoning_tokens": 160,
                "actual_total_tokens": 786242,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 392.4068819950335,
                "first_final_s": 407.99144129501656,
                "completion_s": 413.62771054997575
              },
              {
                "request_id": "flash-capacity-786432-c20-0018",
                "case_id": "capacity-flash-786432-18",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785876,
                "actual_prompt_tokens": 785876,
                "actual_completion_tokens": 352,
                "actual_reasoning_tokens": 159,
                "actual_total_tokens": 786228,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 412.8465921600582,
                "first_final_s": 413.4505691070808,
                "completion_s": 414.3025761710014
              }
            ],
            "effective_timeout_s": {
              "n": 20,
              "median": 1287.969953537162,
              "p95_nearest_rank": 1288.0545065110782,
              "min": 1287.8273294551764,
              "max": 1288.0635407600785
            },
            "deadline_censored": 0
          },
          {
            "condition": "flash-capacity-786432-c5",
            "concurrency": 5,
            "planned": 5,
            "returned": 5,
            "condition_closed": true,
            "not_returned": 0,
            "complete": 5,
            "numeric_citation_pass": 5,
            "statuses": {
              "complete": 5
            },
            "noncompletion_fraction_of_returned": 0.0,
            "completion_fraction_of_planned": 1.0,
            "successful_latency_s": {
              "first_generated_s": {
                "n": 5,
                "min": 22.33168484608177,
                "median": 63.450860679964535,
                "max": 105.82652056997176
              },
              "first_final_s": {
                "n": 5,
                "min": 43.10368414304685,
                "median": 88.80001010990236,
                "max": 106.49563095695339
              },
              "completion_s": {
                "n": 5,
                "min": 58.943790239980444,
                "median": 105.94376130693126,
                "max": 107.23851010994986
              }
            },
            "resources": {
              "sample_count": 11,
              "sampled_gpu_peak_mib_by_index": {
                "0": 151082.0,
                "1": 150460.0,
                "2": 150460.0,
                "3": 150460.0,
                "4": 150460.0,
                "5": 150460.0,
                "6": 150460.0,
                "7": 150460.0
              },
              "minimum_host_memavailable_bytes": 3146050994176,
              "qualification": "10-second samples during the condition; not an exact peak. Host available RAM is host-wide, not model allocation."
            },
            "rows": [
              {
                "request_id": "flash-capacity-786432-c5-0000",
                "case_id": "capacity-flash-786432-00",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785876,
                "actual_prompt_tokens": 785876,
                "actual_completion_tokens": 357,
                "actual_reasoning_tokens": 196,
                "actual_total_tokens": 786233,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 22.33168484608177,
                "first_final_s": 43.10368414304685,
                "completion_s": 58.943790239980444
              },
              {
                "request_id": "flash-capacity-786432-c5-0001",
                "case_id": "capacity-flash-786432-01",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785876,
                "actual_prompt_tokens": 785876,
                "actual_completion_tokens": 311,
                "actual_reasoning_tokens": 141,
                "actual_total_tokens": 786187,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 42.89873219502624,
                "first_final_s": 55.91776136402041,
                "completion_s": 73.61336708406452
              },
              {
                "request_id": "flash-capacity-786432-c5-0002",
                "case_id": "capacity-flash-786432-02",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785876,
                "actual_prompt_tokens": 785876,
                "actual_completion_tokens": 426,
                "actual_reasoning_tokens": 256,
                "actual_total_tokens": 786302,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 63.450860679964535,
                "first_final_s": 88.80001010990236,
                "completion_s": 105.94376130693126
              },
              {
                "request_id": "flash-capacity-786432-c5-0003",
                "case_id": "capacity-flash-786432-03",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785876,
                "actual_prompt_tokens": 785876,
                "actual_completion_tokens": 332,
                "actual_reasoning_tokens": 166,
                "actual_total_tokens": 786208,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 84.02020561590325,
                "first_final_s": 100.49757885094732,
                "completion_s": 106.39020705292933
              },
              {
                "request_id": "flash-capacity-786432-c5-0004",
                "case_id": "capacity-flash-786432-04",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785876,
                "actual_prompt_tokens": 785876,
                "actual_completion_tokens": 338,
                "actual_reasoning_tokens": 168,
                "actual_total_tokens": 786214,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 105.82652056997176,
                "first_final_s": 106.49563095695339,
                "completion_s": 107.23851010994986
              }
            ],
            "effective_timeout_s": {
              "n": 5,
              "median": 1603.8763621819671,
              "p95_nearest_rank": 1603.909062644001,
              "min": 1603.8364100799663,
              "max": 1603.909062644001
            },
            "deadline_censored": 0
          }
        ],
        "largest_complete_input_tokens_c1": 785876,
        "largest_numeric_citation_pass_input_tokens_c1": 785876,
        "largest_all20_complete_input_tokens_lower_bound": 785876,
        "largest_all20_numeric_citation_pass_input_tokens_lower_bound": 785876,
        "customer_latency_SLO": "Not yet agreed; no practical-limit pass/fail inferred.",
        "unmeasured_prompt_targets": [],
        "status": "CAPACITY_RESULTS_AVAILABLE",
        "unmeasured_conditions": []
      },
      "science_grades": [
        {
          "case_id": "reason-05-mixing",
          "request_id": "flash-scientific_reasoning-0004",
          "completion_status": "complete",
          "kind": "numeric",
          "automatic_pass": true,
          "needs_human_review": true,
          "json_format_pass": true,
          "numeric_pass": true,
          "citation_ids_pass": true,
          "parsed_value": 44,
          "expected": 44,
          "human_review_scope": "Units, assumptions, explanation correctness, limitations; automatic numeric pass alone is not full scientific quality."
        },
        {
          "case_id": "reason-02-bayes",
          "request_id": "flash-scientific_reasoning-0001",
          "completion_status": "complete",
          "kind": "numeric",
          "automatic_pass": true,
          "needs_human_review": true,
          "json_format_pass": true,
          "numeric_pass": true,
          "citation_ids_pass": true,
          "parsed_value": 0.1610169492,
          "expected": 0.16101694915254236,
          "human_review_scope": "Units, assumptions, explanation correctness, limitations; automatic numeric pass alone is not full scientific quality."
        },
        {
          "case_id": "reason-06-pump_power",
          "request_id": "flash-scientific_reasoning-0005",
          "completion_status": "complete",
          "kind": "numeric",
          "automatic_pass": true,
          "needs_human_review": true,
          "json_format_pass": true,
          "numeric_pass": true,
          "citation_ids_pass": true,
          "parsed_value": 1000,
          "expected": 1000,
          "human_review_scope": "Units, assumptions, explanation correctness, limitations; automatic numeric pass alone is not full scientific quality."
        },
        {
          "case_id": "reason-04-arrhenius",
          "request_id": "flash-scientific_reasoning-0003",
          "completion_status": "complete",
          "kind": "numeric",
          "automatic_pass": true,
          "needs_human_review": true,
          "json_format_pass": true,
          "numeric_pass": true,
          "citation_ids_pass": true,
          "parsed_value": 30.1,
          "expected": 30.141746474355973,
          "human_review_scope": "Units, assumptions, explanation correctness, limitations; automatic numeric pass alone is not full scientific quality."
        },
        {
          "case_id": "reason-03-weighted_mean",
          "request_id": "flash-scientific_reasoning-0002",
          "completion_status": "complete",
          "kind": "numeric",
          "automatic_pass": true,
          "needs_human_review": true,
          "json_format_pass": true,
          "numeric_pass": true,
          "citation_ids_pass": true,
          "parsed_value": 10.8,
          "expected": 10.8,
          "human_review_scope": "Units, assumptions, explanation correctness, limitations; automatic numeric pass alone is not full scientific quality."
        },
        {
          "case_id": "reason-09-diffusion_stability",
          "request_id": "flash-scientific_reasoning-0008",
          "completion_status": "complete",
          "kind": "numeric",
          "automatic_pass": true,
          "needs_human_review": true,
          "json_format_pass": true,
          "numeric_pass": true,
          "citation_ids_pass": true,
          "parsed_value": 0.05,
          "expected": 0.05,
          "human_review_scope": "Units, assumptions, explanation correctness, limitations; automatic numeric pass alone is not full scientific quality."
        },
        {
          "case_id": "reason-08-grid_order",
          "request_id": "flash-scientific_reasoning-0007",
          "completion_status": "complete",
          "kind": "numeric",
          "automatic_pass": true,
          "needs_human_review": true,
          "json_format_pass": true,
          "numeric_pass": true,
          "citation_ids_pass": true,
          "parsed_value": 2.0,
          "expected": 2,
          "human_review_scope": "Units, assumptions, explanation correctness, limitations; automatic numeric pass alone is not full scientific quality."
        },
        {
          "case_id": "reason-01-uncertainty",
          "request_id": "flash-scientific_reasoning-0000",
          "completion_status": "complete",
          "kind": "numeric",
          "automatic_pass": false,
          "needs_human_review": true,
          "json_format_pass": true,
          "numeric_pass": false,
          "citation_ids_pass": true,
          "parsed_value": 0.036,
          "expected": 0.03577708763999663,
          "human_review_scope": "Units, assumptions, explanation correctness, limitations; automatic numeric pass alone is not full scientific quality."
        },
        {
          "case_id": "reason-07-decay_inverse",
          "request_id": "flash-scientific_reasoning-0006",
          "completion_status": "complete",
          "kind": "numeric",
          "automatic_pass": true,
          "needs_human_review": true,
          "json_format_pass": true,
          "numeric_pass": true,
          "citation_ids_pass": true,
          "parsed_value": 20,
          "expected": 20,
          "human_review_scope": "Units, assumptions, explanation correctness, limitations; automatic numeric pass alone is not full scientific quality."
        },
        {
          "case_id": "reason-12-serial_yield",
          "request_id": "flash-scientific_reasoning-0011",
          "completion_status": "complete",
          "kind": "numeric",
          "automatic_pass": true,
          "needs_human_review": true,
          "json_format_pass": true,
          "numeric_pass": true,
          "citation_ids_pass": true,
          "parsed_value": 0.6,
          "expected": 0.6,
          "human_review_scope": "Units, assumptions, explanation correctness, limitations; automatic numeric pass alone is not full scientific quality."
        },
        {
          "case_id": "reason-10-correlated_uncertainty",
          "request_id": "flash-scientific_reasoning-0009",
          "completion_status": "complete",
          "kind": "numeric",
          "automatic_pass": true,
          "needs_human_review": true,
          "json_format_pass": true,
          "numeric_pass": true,
          "citation_ids_pass": true,
          "parsed_value": 0.632455532,
          "expected": 0.6324555320336759,
          "human_review_scope": "Units, assumptions, explanation correctness, limitations; automatic numeric pass alone is not full scientific quality."
        },
        {
          "case_id": "reason-11-unit_conversion",
          "request_id": "flash-scientific_reasoning-0010",
          "completion_status": "complete",
          "kind": "numeric",
          "automatic_pass": true,
          "needs_human_review": true,
          "json_format_pass": true,
          "numeric_pass": true,
          "citation_ids_pass": true,
          "parsed_value": 192,
          "expected": 192,
          "human_review_scope": "Units, assumptions, explanation correctness, limitations; automatic numeric pass alone is not full scientific quality."
        },
        {
          "case_id": "reason-13-confounding",
          "request_id": "flash-scientific_reasoning-0012",
          "completion_status": "complete",
          "kind": "human_rubric",
          "automatic_pass": null,
          "needs_human_review": true,
          "rubric": [
            "材料と温度が交絡し因果効果は未同定",
            "同温度比較または材料×温度の要因実験",
            "無作為化・反復・不確かさ"
          ],
          "json_format_pass": true
        },
        {
          "case_id": "reason-14-missing_heat",
          "request_id": "flash-scientific_reasoning-0013",
          "completion_status": "complete",
          "kind": "human_rubric",
          "automatic_pass": null,
          "needs_human_review": true,
          "rubric": [
            "温度差だけから熱量は決まらない",
            "質量・比熱・容器熱容量が必要",
            "熱損失や外部加熱・反応量と符号の確認"
          ],
          "json_format_pass": true
        },
        {
          "case_id": "reason-15-pvalue",
          "request_id": "flash-scientific_reasoning-0014",
          "completion_status": "complete",
          "kind": "human_rubric",
          "automatic_pass": null,
          "needs_human_review": true,
          "rubric": [
            "p値は仮説の事後確率ではない",
            "帰無仮説とモデルの下でのデータの極端さ",
            "効果量・区間・多重比較・事前計画が必要"
          ],
          "json_format_pass": true
        },
        {
          "case_id": "reason-20-identifiability",
          "request_id": "flash-scientific_reasoning-0019",
          "completion_status": "complete",
          "kind": "human_rubric",
          "automatic_pass": null,
          "needs_human_review": true,
          "rubric": [
            "識別できるのは積ab=6のみ",
            "(a,b)の複数の組が同じ観測を生成",
            "片方の独立測定または異なる依存性の観測が必要"
          ],
          "json_format_pass": true
        },
        {
          "case_id": "reason-18-overfitting",
          "request_id": "flash-scientific_reasoning-0017",
          "completion_status": "complete",
          "kind": "human_rubric",
          "automatic_pass": null,
          "needs_human_review": true,
          "rubric": [
            "特徴選択と前処理で検証側情報が漏れる",
            "各訓練fold内で前処理と特徴選択をfit",
            "調整にはnested CVまたは独立test"
          ],
          "json_format_pass": true
        },
        {
          "case_id": "reason-16-simpson",
          "request_id": "flash-scientific_reasoning-0015",
          "completion_status": "complete",
          "kind": "human_rubric",
          "automatic_pass": null,
          "needs_human_review": true,
          "rubric": [
            "A全体83% B全体27%",
            "簡単条件は両方90%、困難条件は両方20%",
            "条件構成の違いで全体差が生じる"
          ],
          "json_format_pass": true
        },
        {
          "case_id": "reason-17-negative_control",
          "request_id": "flash-scientific_reasoning-0016",
          "completion_status": "complete",
          "kind": "human_rubric",
          "automatic_pass": null,
          "needs_human_review": true,
          "rubric": [
            "温度と撹拌を統制し実測",
            "触媒なし対照および適切な担体対照",
            "独立反復・無作為化・速度の定量と不確かさ"
          ],
          "json_format_pass": true
        },
        {
          "case_id": "reason-19-falsification",
          "request_id": "flash-scientific_reasoning-0018",
          "completion_status": "complete",
          "kind": "human_rubric",
          "automatic_pass": null,
          "needs_human_review": true,
          "rubric": [
            "反証可能な事前予測が不足",
            "比較する代替仮説と測定量・許容範囲",
            "観測後の恣意的な条件変更を避ける"
          ],
          "json_format_pass": true
        }
      ]
    },
    "glm": {
      "model": {
        "tag": "glm",
        "repo": "zai-org/GLM-5.3",
        "revision": "ba32f8c19c8738e2c57c71798c601d477a7bf844",
        "profiles": {
          "short_science_qa": {
            "max_tokens": 262144,
            "timeout_s": 5400,
            "chat_template_kwargs": {
              "reasoning_effort": "low"
            },
            "temperature": 1.0,
            "top_p": 0.95
          },
          "long_document_grounded_qa": {
            "max_tokens": 262144,
            "timeout_s": 5400,
            "chat_template_kwargs": {
              "reasoning_effort": "low"
            },
            "temperature": 1.0,
            "top_p": 0.95
          },
          "scientific_reasoning": {
            "max_tokens": 262144,
            "timeout_s": 5400,
            "chat_template_kwargs": {
              "reasoning_effort": "max"
            },
            "temperature": 1.0,
            "top_p": 0.95
          },
          "scientific_code": {
            "max_tokens": 262144,
            "timeout_s": 5400,
            "chat_template_kwargs": {
              "reasoning_effort": "max"
            },
            "temperature": 1.0,
            "top_p": 0.95
          },
          "gpqa_diamond": {
            "max_tokens": 262144,
            "timeout_s": 5400,
            "chat_template_kwargs": {
              "reasoning_effort": "max"
            },
            "temperature": 1.0,
            "top_p": 0.95
          }
        }
      },
      "context_tokens": 1048576,
      "max_output_tokens": 262144,
      "gpu_memory_fraction": 0.82,
      "gpqa": {
        "condition": "glm-gpqa_diamond",
        "concurrency": 20,
        "planned": 198,
        "started": 198,
        "complete": 196,
        "statuses": {
          "complete": 196,
          "truncated": 2
        },
        "automatic_pass": 182,
        "automatic_fail": 14,
        "rubric_only": 0,
        "unscored_incomplete": 2,
        "latency_s_complete_only": {
          "first_generated_s": {
            "n": 196,
            "median": 0.11446551751578227,
            "p95_nearest_rank": 1.223112409003079,
            "min": 0.0975425069918856,
            "max": 2.329526312998496
          },
          "first_final_s": {
            "n": 196,
            "median": 120.71199711656664,
            "p95_nearest_rank": 952.2375360820442,
            "min": 1.7102437219582498,
            "max": 2184.8802265180275
          },
          "completion_s": {
            "n": 196,
            "median": 123.38053340348415,
            "p95_nearest_rank": 954.1861565430881,
            "min": 3.07137923699338,
            "max": 2186.903144691023
          }
        },
        "elapsed_s": 4593.72803721705,
        "actual_prompt_tokens": {
          "n": 198,
          "median": 239.5,
          "p95_nearest_rank": 490,
          "min": 105,
          "max": 2830
        },
        "actual_completion_tokens": {
          "n": 198,
          "median": 7860.5,
          "p95_nearest_rank": 63922,
          "min": 182,
          "max": 262144
        },
        "effective_timeout_s": {
          "n": 198,
          "median": 5400.0,
          "p95_nearest_rank": 5400,
          "min": 5400,
          "max": 5400
        },
        "deadline_censored": 0,
        "requests_sha256": "684aeffbeacfb3f2eeca4a80687b1ecd772ffe2135c5a0137ed4ca30f8ac5946",
        "profile": {
          "max_tokens": 262144,
          "timeout_s": 5400,
          "chat_template_kwargs": {
            "reasoning_effort": "max"
          },
          "temperature": 1.0,
          "top_p": 0.95
        },
        "correct_fraction_of_all_198": 0.9191919191919192,
        "correct_fraction_of_completed": 0.9285714285714286
      },
      "science": {
        "condition": "glm-scientific_reasoning",
        "concurrency": 5,
        "planned": 20,
        "started": 20,
        "complete": 20,
        "statuses": {
          "complete": 20
        },
        "automatic_pass": 10,
        "automatic_fail": 2,
        "rubric_only": 8,
        "unscored_incomplete": 0,
        "latency_s_complete_only": {
          "first_generated_s": {
            "n": 20,
            "median": 0.10740695905406028,
            "p95_nearest_rank": 0.2865337379043922,
            "min": 0.09504049294628203,
            "max": 0.2867314749164507
          },
          "first_final_s": {
            "n": 20,
            "median": 7.09836300247116,
            "p95_nearest_rank": 50.654020059038885,
            "min": 1.3736677799606696,
            "max": 68.57553677202668
          },
          "completion_s": {
            "n": 20,
            "median": 11.409728199942037,
            "p95_nearest_rank": 59.248965515056625,
            "min": 3.635636919992976,
            "max": 77.34131439903285
          }
        },
        "elapsed_s": 110.12084570794832,
        "actual_prompt_tokens": {
          "n": 20,
          "median": 199.5,
          "p95_nearest_rank": 215,
          "min": 184,
          "max": 220
        },
        "actual_completion_tokens": {
          "n": 20,
          "median": 972.5,
          "p95_nearest_rank": 5497,
          "min": 316,
          "max": 7790
        },
        "effective_timeout_s": {
          "n": 20,
          "median": 5387.93031563994,
          "p95_nearest_rank": 5399.99873709993,
          "min": 5346.599190522917,
          "max": 5399.999019664945
        },
        "deadline_censored": 0,
        "requests_sha256": "45e107a2188814e7e754077713b42dca04d19f1c60b992b16cbce2f4f13f10ad",
        "profile": {
          "max_tokens": 262144,
          "timeout_s": 5400,
          "chat_template_kwargs": {
            "reasoning_effort": "max"
          },
          "temperature": 1.0,
          "top_p": 0.95
        }
      },
      "code": {
        "generated": 10,
        "task_pass": 9,
        "task_fail": 1,
        "fixed_tests": 61,
        "test_pass": 60,
        "failed_cases": [
          {
            "request_id": "glm-scientific_code-0006",
            "case_id": "code-07-confusion_metrics",
            "completion_status": "complete",
            "passed": false,
            "returncode": 0,
            "stdout": "{\"passed\": false, \"tests\": [{\"pass\": true, \"actual\": {\"precision\": 0.8, \"recall\": 0.6666666666666666, \"specificity\": 0.75, \"f1\": 0.7272727272727272}, \"expected\": {\"precision\": 0.8, \"recall\": 0.6666666666666666, \"specificity\": 0.75, \"f1\": 0.7272727272727273}}, {\"pass\": true, \"actual\": {\"precision\": null, \"recall\": null, \"specificity\": null, \"f1\": null}, \"expected\": {\"precision\": null, \"recall\": null, \"specificity\": null, \"f1\": null}}, {\"pass\": false, \"actual\": {\"precision\": null, \"recall\": 0.0, \"specificity\": 1.0, \"f1\": null}, \"expected\": {\"precision\": null, \"recall\": 0, \"specificity\": 1, \"f1\": 0}}, {\"pass\": true, \"actual\": {\"precision\": 1.0, \"recall\": 1.0, \"specificity\": 1.0, \"f1\": 1.0}, \"expected\": {\"precision\": 1, \"recall\": 1, \"specificity\": 1, \"f1\": 1}}, {\"pass\": true, \"actual\": null, \"expected\": null}, {\"pass\": true, \"actual\": null, \"expected\": null}]}\n",
            "stderr": "",
            "elapsed_s": 0.14058445300906897,
            "judgement": {
              "passed": false,
              "tests": [
                {
                  "pass": true,
                  "actual": {
                    "precision": 0.8,
                    "recall": 0.6666666666666666,
                    "specificity": 0.75,
                    "f1": 0.7272727272727272
                  },
                  "expected": {
                    "precision": 0.8,
                    "recall": 0.6666666666666666,
                    "specificity": 0.75,
                    "f1": 0.7272727272727273
                  }
                },
                {
                  "pass": true,
                  "actual": {
                    "precision": null,
                    "recall": null,
                    "specificity": null,
                    "f1": null
                  },
                  "expected": {
                    "precision": null,
                    "recall": null,
                    "specificity": null,
                    "f1": null
                  }
                },
                {
                  "pass": false,
                  "actual": {
                    "precision": null,
                    "recall": 0.0,
                    "specificity": 1.0,
                    "f1": null
                  },
                  "expected": {
                    "precision": null,
                    "recall": 0,
                    "specificity": 1,
                    "f1": 0
                  }
                },
                {
                  "pass": true,
                  "actual": {
                    "precision": 1.0,
                    "recall": 1.0,
                    "specificity": 1.0,
                    "f1": 1.0
                  },
                  "expected": {
                    "precision": 1,
                    "recall": 1,
                    "specificity": 1,
                    "f1": 1
                  }
                },
                {
                  "pass": true,
                  "actual": null,
                  "expected": null
                },
                {
                  "pass": true,
                  "actual": null,
                  "expected": null
                }
              ]
            }
          }
        ]
      },
      "operational": [
        {
          "condition": "glm-fair262k-c1-long_document_grounded_qa",
          "concurrency": 1,
          "planned": 20,
          "started": 20,
          "complete": 20,
          "statuses": {
            "complete": 20
          },
          "automatic_pass": 19,
          "automatic_fail": 1,
          "rubric_only": 0,
          "unscored_incomplete": 0,
          "latency_s_complete_only": {
            "first_generated_s": {
              "n": 20,
              "median": 0.8053059609956108,
              "p95_nearest_rank": 0.8112973809475079,
              "min": 0.8031814660644159,
              "max": 0.8179717110469937
            },
            "first_final_s": {
              "n": 20,
              "median": 0.9779042055015452,
              "p95_nearest_rank": 1.1363220389466733,
              "min": 0.8106809130404145,
              "max": 1.137057923944667
            },
            "completion_s": {
              "n": 20,
              "median": 2.655324879975524,
              "p95_nearest_rank": 3.008222009986639,
              "min": 2.250007251976058,
              "max": 3.165838909917511
            }
          },
          "elapsed_s": 52.972097750985995,
          "actual_prompt_tokens": {
            "n": 20,
            "median": 17835.5,
            "p95_nearest_rank": 17848,
            "min": 17811,
            "max": 17849
          },
          "actual_completion_tokens": {
            "n": 20,
            "median": 211.0,
            "p95_nearest_rank": 251,
            "min": 165,
            "max": 269
          },
          "effective_timeout_s": {
            "n": 20,
            "median": 1744.8549032733426,
            "p95_nearest_rank": 1766.9717580159195,
            "min": 1719.1975105738966,
            "max": 1769.4437561678933
          },
          "deadline_censored": 0,
          "requests_sha256": "c30f797089b5e78304cd6c655f89cfec7f98b509c706c3ba734b1c899bb7c31b",
          "profile": {
            "max_tokens": 262144,
            "timeout_s": 5400,
            "chat_template_kwargs": {
              "reasoning_effort": "low"
            },
            "temperature": 1.0,
            "top_p": 0.95
          }
        },
        {
          "condition": "glm-fair262k-c1-scientific_reasoning",
          "concurrency": 1,
          "planned": 20,
          "started": 20,
          "complete": 20,
          "statuses": {
            "complete": 20
          },
          "automatic_pass": 11,
          "automatic_fail": 1,
          "rubric_only": 8,
          "unscored_incomplete": 0,
          "latency_s_complete_only": {
            "first_generated_s": {
              "n": 20,
              "median": 0.08312511845724657,
              "p95_nearest_rank": 0.08875833288766444,
              "min": 0.08098872494883835,
              "max": 0.09098394110333174
            },
            "first_final_s": {
              "n": 20,
              "median": 6.195475970511325,
              "p95_nearest_rank": 61.55289802199695,
              "min": 1.1240539179416373,
              "max": 65.49730760999955
            },
            "completion_s": {
              "n": 20,
              "median": 9.044314576021861,
              "p95_nearest_rank": 68.45510258106515,
              "min": 3.0985547859454527,
              "max": 73.84207955794409
            }
          },
          "elapsed_s": 359.51973266503774,
          "actual_prompt_tokens": {
            "n": 20,
            "median": 199.5,
            "p95_nearest_rank": 215,
            "min": 184,
            "max": 220
          },
          "actual_completion_tokens": {
            "n": 20,
            "median": 1069.5,
            "p95_nearest_rank": 7982,
            "min": 365,
            "max": 8604
          },
          "effective_timeout_s": {
            "n": 20,
            "median": 1648.1258730319678,
            "p95_nearest_rank": 1709.0168243739754,
            "min": 1374.281319922884,
            "max": 1716.4699701858917
          },
          "deadline_censored": 0,
          "requests_sha256": "8edb9038c5945d2fe6a665d91af443771428491468e9b2d320b7492faa20b8f5",
          "profile": {
            "max_tokens": 262144,
            "timeout_s": 5400,
            "chat_template_kwargs": {
              "reasoning_effort": "max"
            },
            "temperature": 1.0,
            "top_p": 0.95
          }
        },
        {
          "condition": "glm-fair262k-c1-short_science_qa",
          "concurrency": 1,
          "planned": 20,
          "started": 20,
          "complete": 20,
          "statuses": {
            "complete": 20
          },
          "automatic_pass": 18,
          "automatic_fail": 2,
          "rubric_only": 0,
          "unscored_incomplete": 0,
          "latency_s_complete_only": {
            "first_generated_s": {
              "n": 20,
              "median": 0.08947076956974342,
              "p95_nearest_rank": 0.09198643802665174,
              "min": 0.0802640609908849,
              "max": 0.09713732497766614
            },
            "first_final_s": {
              "n": 20,
              "median": 0.0896653970121406,
              "p95_nearest_rank": 0.2701649700757116,
              "min": 0.08859507297165692,
              "max": 0.5828283750452101
            },
            "completion_s": {
              "n": 20,
              "median": 1.3149173440178856,
              "p95_nearest_rank": 2.395550201064907,
              "min": 0.9708296019816771,
              "max": 2.6774069279199466
            }
          },
          "elapsed_s": 30.55408233695198,
          "actual_prompt_tokens": {
            "n": 20,
            "median": 182.5,
            "p95_nearest_rank": 190,
            "min": 172,
            "max": 194
          },
          "actual_completion_tokens": {
            "n": 20,
            "median": 150.5,
            "p95_nearest_rank": 281,
            "min": 108,
            "max": 315
          },
          "effective_timeout_s": {
            "n": 20,
            "median": 1785.9545431344886,
            "p95_nearest_rank": 1799.0277410330018,
            "min": 1770.6470950939693,
            "max": 1799.9995595780201
          },
          "deadline_censored": 0,
          "requests_sha256": "d59a3acaf4d41e0111dd49f91fad06984db2ad3a589eceac1b44f601c0449ff8",
          "profile": {
            "max_tokens": 262144,
            "timeout_s": 5400,
            "chat_template_kwargs": {
              "reasoning_effort": "low"
            },
            "temperature": 1.0,
            "top_p": 0.95
          }
        },
        {
          "condition": "glm-fair262k-c10-long_document_grounded_qa",
          "concurrency": 10,
          "planned": 20,
          "started": 20,
          "complete": 20,
          "statuses": {
            "complete": 20
          },
          "automatic_pass": 20,
          "automatic_fail": 0,
          "rubric_only": 0,
          "unscored_incomplete": 0,
          "latency_s_complete_only": {
            "first_generated_s": {
              "n": 20,
              "median": 1.6021577029605396,
              "p95_nearest_rank": 7.042308849981055,
              "min": 0.8304566210135818,
              "max": 7.701635302975774
            },
            "first_final_s": {
              "n": 20,
              "median": 4.244003522500861,
              "p95_nearest_rank": 7.9441705900244415,
              "min": 1.7590808420209214,
              "max": 8.022276214906015
            },
            "completion_s": {
              "n": 20,
              "median": 9.919736970507074,
              "p95_nearest_rank": 16.609175917925313,
              "min": 3.8395881300093606,
              "max": 16.64883767394349
            }
          },
          "elapsed_s": 21.038914423901588,
          "actual_prompt_tokens": {
            "n": 20,
            "median": 17835.5,
            "p95_nearest_rank": 17848,
            "min": 17811,
            "max": 17849
          },
          "actual_completion_tokens": {
            "n": 20,
            "median": 213.0,
            "p95_nearest_rank": 268,
            "min": 166,
            "max": 278
          },
          "effective_timeout_s": {
            "n": 20,
            "median": 1198.0317183354637,
            "p95_nearest_rank": 1202.900989637943,
            "min": 1186.2508766449755,
            "max": 1202.9012931329198
          },
          "deadline_censored": 0,
          "requests_sha256": "006055a18438d8a8dfef7be3d9a0eb0ec18d42196c239529d483439b42e5539e",
          "profile": {
            "max_tokens": 262144,
            "timeout_s": 5400,
            "chat_template_kwargs": {
              "reasoning_effort": "low"
            },
            "temperature": 1.0,
            "top_p": 0.95
          }
        },
        {
          "condition": "glm-fair262k-c10-scientific_reasoning",
          "concurrency": 10,
          "planned": 20,
          "started": 20,
          "complete": 20,
          "statuses": {
            "complete": 20
          },
          "automatic_pass": 12,
          "automatic_fail": 0,
          "rubric_only": 8,
          "unscored_incomplete": 0,
          "latency_s_complete_only": {
            "first_generated_s": {
              "n": 20,
              "median": 0.10836900846334174,
              "p95_nearest_rank": 0.19011063303332776,
              "min": 0.0862355490680784,
              "max": 0.19015290203969926
            },
            "first_final_s": {
              "n": 20,
              "median": 6.89242602349259,
              "p95_nearest_rank": 45.741871702019125,
              "min": 2.011767375981435,
              "max": 62.02254902396817
            },
            "completion_s": {
              "n": 20,
              "median": 13.064313150534872,
              "p95_nearest_rank": 54.63051089295186,
              "min": 4.858911190996878,
              "max": 70.25862582493573
            }
          },
          "elapsed_s": 81.97168191301171,
          "actual_prompt_tokens": {
            "n": 20,
            "median": 199.5,
            "p95_nearest_rank": 215,
            "min": 184,
            "max": 220
          },
          "actual_completion_tokens": {
            "n": 20,
            "median": 964.0,
            "p95_nearest_rank": 5235,
            "min": 355,
            "max": 6988
          },
          "effective_timeout_s": {
            "n": 20,
            "median": 1179.1519556815038,
            "p95_nearest_rank": 1181.8595985360444,
            "min": 1168.694804484956,
            "max": 1181.8598638420226
          },
          "deadline_censored": 0,
          "requests_sha256": "0ea7f1b80eb0b89def0ee9803c55bc4489b2d90691c4b5e30436c0e6741d0c6c",
          "profile": {
            "max_tokens": 262144,
            "timeout_s": 5400,
            "chat_template_kwargs": {
              "reasoning_effort": "max"
            },
            "temperature": 1.0,
            "top_p": 0.95
          }
        },
        {
          "condition": "glm-fair262k-c10-short_science_qa",
          "concurrency": 10,
          "planned": 20,
          "started": 20,
          "complete": 20,
          "statuses": {
            "complete": 20
          },
          "automatic_pass": 18,
          "automatic_fail": 2,
          "rubric_only": 0,
          "unscored_incomplete": 0,
          "latency_s_complete_only": {
            "first_generated_s": {
              "n": 20,
              "median": 0.19312365853693336,
              "p95_nearest_rank": 0.2088357510510832,
              "min": 0.10064108902588487,
              "max": 0.20917413593269885
            },
            "first_final_s": {
              "n": 20,
              "median": 0.20457902742782608,
              "p95_nearest_rank": 0.5466623170068488,
              "min": 0.10064108902588487,
              "max": 0.8056584500009194
            },
            "completion_s": {
              "n": 20,
              "median": 2.4591089515015483,
              "p95_nearest_rank": 4.506200293079019,
              "min": 1.344022385077551,
              "max": 4.713204356958158
            }
          },
          "elapsed_s": 7.2106560110114515,
          "actual_prompt_tokens": {
            "n": 20,
            "median": 182.5,
            "p95_nearest_rank": 190,
            "min": 172,
            "max": 194
          },
          "actual_completion_tokens": {
            "n": 20,
            "median": 158.5,
            "p95_nearest_rank": 308,
            "min": 97,
            "max": 417
          },
          "effective_timeout_s": {
            "n": 20,
            "median": 1209.2638469014782,
            "p95_nearest_rank": 1210.114286973956,
            "min": 1205.5434902269626,
            "max": 1210.1145658389432
          },
          "deadline_censored": 0,
          "requests_sha256": "8adbff066bbf39f1589912d3b3be23b00a77211ab50f5a310cc4162ab8c34257",
          "profile": {
            "max_tokens": 262144,
            "timeout_s": 5400,
            "chat_template_kwargs": {
              "reasoning_effort": "low"
            },
            "temperature": 1.0,
            "top_p": 0.95
          }
        },
        {
          "condition": "glm-fair262k-c20-long_document_grounded_qa",
          "concurrency": 20,
          "planned": 20,
          "started": 20,
          "complete": 20,
          "statuses": {
            "complete": 20
          },
          "automatic_pass": 19,
          "automatic_fail": 1,
          "rubric_only": 0,
          "unscored_incomplete": 0,
          "latency_s_complete_only": {
            "first_generated_s": {
              "n": 20,
              "median": 8.115217046521138,
              "p95_nearest_rank": 14.627252261037938,
              "min": 0.9102059700526297,
              "max": 15.419451612979174
            },
            "first_final_s": {
              "n": 20,
              "median": 12.416893488029018,
              "p95_nearest_rank": 15.696549984975718,
              "min": 1.781343883019872,
              "max": 15.707331974990666
            },
            "completion_s": {
              "n": 20,
              "median": 17.87715112499427,
              "p95_nearest_rank": 18.605168865062296,
              "min": 17.168562048929743,
              "max": 19.535411286051385
            }
          },
          "elapsed_s": 19.53678490302991,
          "actual_prompt_tokens": {
            "n": 20,
            "median": 17835.5,
            "p95_nearest_rank": 17848,
            "min": 17811,
            "max": 17849
          },
          "actual_completion_tokens": {
            "n": 20,
            "median": 208.5,
            "p95_nearest_rank": 246,
            "min": 177,
            "max": 421
          },
          "effective_timeout_s": {
            "n": 20,
            "median": 1095.3621541385073,
            "p95_nearest_rank": 1095.3643512519775,
            "min": 1095.3598610570189,
            "max": 1095.3646560270572
          },
          "deadline_censored": 0,
          "requests_sha256": "b71e076189c19bf5181587ebb8806b8b002faa757a95dcf075ceec3426f2e342",
          "profile": {
            "max_tokens": 262144,
            "timeout_s": 5400,
            "chat_template_kwargs": {
              "reasoning_effort": "low"
            },
            "temperature": 1.0,
            "top_p": 0.95
          }
        },
        {
          "condition": "glm-fair262k-c20-scientific_reasoning",
          "concurrency": 20,
          "planned": 20,
          "started": 20,
          "complete": 20,
          "statuses": {
            "complete": 20
          },
          "automatic_pass": 11,
          "automatic_fail": 1,
          "rubric_only": 8,
          "unscored_incomplete": 0,
          "latency_s_complete_only": {
            "first_generated_s": {
              "n": 20,
              "median": 0.22311752097448334,
              "p95_nearest_rank": 0.22415259899571538,
              "min": 0.0940844650613144,
              "max": 0.22427110595162958
            },
            "first_final_s": {
              "n": 20,
              "median": 8.467174403485842,
              "p95_nearest_rank": 39.24499114998616,
              "min": 1.8707527989754453,
              "max": 67.56364439590834
            },
            "completion_s": {
              "n": 20,
              "median": 14.061435785493813,
              "p95_nearest_rank": 49.213478544028476,
              "min": 4.409369718981907,
              "max": 72.29915442597121
            }
          },
          "elapsed_s": 72.30101088504307,
          "actual_prompt_tokens": {
            "n": 20,
            "median": 199.5,
            "p95_nearest_rank": 215,
            "min": 184,
            "max": 220
          },
          "actual_completion_tokens": {
            "n": 20,
            "median": 969.0,
            "p95_nearest_rank": 4375,
            "min": 284,
            "max": 7052
          },
          "effective_timeout_s": {
            "n": 20,
            "median": 1075.822193269385,
            "p95_nearest_rank": 1075.8233422398334,
            "min": 1075.8208273618948,
            "max": 1075.8236264148727
          },
          "deadline_censored": 0,
          "requests_sha256": "f6ff55309ae34b2869236a1448822b728f0d99fc94f980ee3ee0c0bc6d6b8b8d",
          "profile": {
            "max_tokens": 262144,
            "timeout_s": 5400,
            "chat_template_kwargs": {
              "reasoning_effort": "max"
            },
            "temperature": 1.0,
            "top_p": 0.95
          }
        },
        {
          "condition": "glm-fair262k-c20-short_science_qa",
          "concurrency": 20,
          "planned": 20,
          "started": 20,
          "complete": 20,
          "statuses": {
            "complete": 20
          },
          "automatic_pass": 18,
          "automatic_fail": 2,
          "rubric_only": 0,
          "unscored_incomplete": 0,
          "latency_s_complete_only": {
            "first_generated_s": {
              "n": 20,
              "median": 0.23114621749846265,
              "p95_nearest_rank": 0.23269876604899764,
              "min": 0.11019432602915913,
              "max": 0.23281141405459493
            },
            "first_final_s": {
              "n": 20,
              "median": 0.23159599001519382,
              "p95_nearest_rank": 0.5334711639443412,
              "min": 0.11019432602915913,
              "max": 1.0730813150294125
            },
            "completion_s": {
              "n": 20,
              "median": 2.5980550035019405,
              "p95_nearest_rank": 4.454065793077461,
              "min": 1.7091614020755515,
              "max": 4.514325515017845
            }
          },
          "elapsed_s": 4.515418526018038,
          "actual_prompt_tokens": {
            "n": 20,
            "median": 182.5,
            "p95_nearest_rank": 190,
            "min": 172,
            "max": 194
          },
          "actual_completion_tokens": {
            "n": 20,
            "median": 171.0,
            "p95_nearest_rank": 357,
            "min": 106,
            "max": 364
          },
          "effective_timeout_s": {
            "n": 20,
            "median": 1099.8823611004627,
            "p95_nearest_rank": 1099.8836654740153,
            "min": 1099.8800932730082,
            "max": 1099.8838823999977
          },
          "deadline_censored": 0,
          "requests_sha256": "2883a8fae16bd5d1e3769a79916d00835d8105cf7fc6d54ce716f275ae444f94",
          "profile": {
            "max_tokens": 262144,
            "timeout_s": 5400,
            "chat_template_kwargs": {
              "reasoning_effort": "low"
            },
            "temperature": 1.0,
            "top_p": 0.95
          }
        },
        {
          "condition": "glm-fair262k-c5-long_document_grounded_qa",
          "concurrency": 5,
          "planned": 20,
          "started": 20,
          "complete": 20,
          "statuses": {
            "complete": 20
          },
          "automatic_pass": 19,
          "automatic_fail": 1,
          "rubric_only": 0,
          "unscored_incomplete": 0,
          "latency_s_complete_only": {
            "first_generated_s": {
              "n": 20,
              "median": 0.8259391925530508,
              "p95_nearest_rank": 3.174546162947081,
              "min": 0.8198952539823949,
              "max": 3.8599217699375004
            },
            "first_final_s": {
              "n": 20,
              "median": 1.7139313379884697,
              "p95_nearest_rank": 4.119011331931688,
              "min": 0.8277919030515477,
              "max": 4.291825621970929
            },
            "completion_s": {
              "n": 20,
              "median": 6.151019945566077,
              "p95_nearest_rank": 8.411988867912441,
              "min": 3.1077768630348146,
              "max": 9.668011561967432
            }
          },
          "elapsed_s": 25.375165123026818,
          "actual_prompt_tokens": {
            "n": 20,
            "median": 17835.5,
            "p95_nearest_rank": 17848,
            "min": 17811,
            "max": 17849
          },
          "actual_completion_tokens": {
            "n": 20,
            "median": 208.0,
            "p95_nearest_rank": 255,
            "min": 166,
            "max": 312
          },
          "effective_timeout_s": {
            "n": 20,
            "median": 1336.1746706525446,
            "p95_nearest_rank": 1346.8788966169814,
            "min": 1324.6118010220816,
            "max": 1346.8792045520386
          },
          "deadline_censored": 0,
          "requests_sha256": "d0f338bbc07678987a64fbfdd9fe208fa92a8a2fb3018c1eb83f19fb01644a0e",
          "profile": {
            "max_tokens": 262144,
            "timeout_s": 5400,
            "chat_template_kwargs": {
              "reasoning_effort": "low"
            },
            "temperature": 1.0,
            "top_p": 0.95
          }
        },
        {
          "condition": "glm-fair262k-c5-scientific_reasoning",
          "concurrency": 5,
          "planned": 20,
          "started": 20,
          "complete": 20,
          "statuses": {
            "complete": 20
          },
          "automatic_pass": 12,
          "automatic_fail": 0,
          "rubric_only": 8,
          "unscored_incomplete": 0,
          "latency_s_complete_only": {
            "first_generated_s": {
              "n": 20,
              "median": 0.10751061199698597,
              "p95_nearest_rank": 1.477903606952168,
              "min": 0.08124216494616121,
              "max": 1.4781383930239826
            },
            "first_final_s": {
              "n": 20,
              "median": 7.9409920735051855,
              "p95_nearest_rank": 59.25610875093844,
              "min": 1.781635291990824,
              "max": 81.92563282395713
            },
            "completion_s": {
              "n": 20,
              "median": 12.671080571017228,
              "p95_nearest_rank": 66.74575074203312,
              "min": 3.986144240014255,
              "max": 88.82718043099158
            }
          },
          "elapsed_s": 111.38437568303198,
          "actual_prompt_tokens": {
            "n": 20,
            "median": 199.5,
            "p95_nearest_rank": 215,
            "min": 184,
            "max": 220
          },
          "actual_completion_tokens": {
            "n": 20,
            "median": 1062.5,
            "p95_nearest_rank": 6434,
            "min": 332,
            "max": 8451
          },
          "effective_timeout_s": {
            "n": 20,
            "median": 1304.1793586058775,
            "p95_nearest_rank": 1321.501769355964,
            "min": 1265.811264875927,
            "max": 1321.5020186419133
          },
          "deadline_censored": 0,
          "requests_sha256": "3e1b6188c3042914f09758500738d04a6822468890cc77e99fc866687d9bc6dd",
          "profile": {
            "max_tokens": 262144,
            "timeout_s": 5400,
            "chat_template_kwargs": {
              "reasoning_effort": "max"
            },
            "temperature": 1.0,
            "top_p": 0.95
          }
        },
        {
          "condition": "glm-fair262k-c5-short_science_qa",
          "concurrency": 5,
          "planned": 20,
          "started": 20,
          "complete": 20,
          "statuses": {
            "complete": 20
          },
          "automatic_pass": 18,
          "automatic_fail": 2,
          "rubric_only": 0,
          "unscored_incomplete": 0,
          "latency_s_complete_only": {
            "first_generated_s": {
              "n": 20,
              "median": 0.1171531134750694,
              "p95_nearest_rank": 0.20452793198637664,
              "min": 0.09455182403326035,
              "max": 0.20472806901670992
            },
            "first_final_s": {
              "n": 20,
              "median": 0.11723883700324222,
              "p95_nearest_rank": 0.3817232290748507,
              "min": 0.09455182403326035,
              "max": 0.5518458139849827
            },
            "completion_s": {
              "n": 20,
              "median": 2.155462148541119,
              "p95_nearest_rank": 3.35147215006873,
              "min": 1.121998441987671,
              "max": 4.97972847300116
            }
          },
          "elapsed_s": 10.066758900065906,
          "actual_prompt_tokens": {
            "n": 20,
            "median": 182.5,
            "p95_nearest_rank": 190,
            "min": 172,
            "max": 194
          },
          "actual_completion_tokens": {
            "n": 20,
            "median": 165.5,
            "p95_nearest_rank": 299,
            "min": 93,
            "max": 384
          },
          "effective_timeout_s": {
            "n": 20,
            "median": 1353.3433754549478,
            "p95_nearest_rank": 1356.947748391889,
            "min": 1348.951507865917,
            "max": 1356.9480246669846
          },
          "deadline_censored": 0,
          "requests_sha256": "e3a6850c3be6ab2970d3c358cadf6d6ef11cca92eef3968b4870bbdd310196a6",
          "profile": {
            "max_tokens": 262144,
            "timeout_s": 5400,
            "chat_template_kwargs": {
              "reasoning_effort": "low"
            },
            "temperature": 1.0,
            "top_p": 0.95
          }
        }
      ],
      "capacity": {
        "repo": "zai-org/GLM-5.3",
        "revision": "ba32f8c19c8738e2c57c71798c601d477a7bf844",
        "configured_context_tokens": 1048576,
        "startup": {
          "ready": true,
          "seconds": 362.16841373208445,
          "cumulative_seconds": 2177.5259957413655,
          "returncode": null
        },
        "benchmark_error": null,
        "planned_prompt_targets": [
          131072,
          262144,
          524288,
          786432
        ],
        "conditions": [
          {
            "condition": "glm-capacity-131072-c1",
            "concurrency": 1,
            "planned": 1,
            "returned": 1,
            "condition_closed": true,
            "not_returned": 0,
            "complete": 1,
            "numeric_citation_pass": 1,
            "statuses": {
              "complete": 1
            },
            "noncompletion_fraction_of_returned": 0.0,
            "completion_fraction_of_planned": 1.0,
            "successful_latency_s": {
              "first_generated_s": {
                "n": 1,
                "min": 7.009602441918105,
                "median": 7.009602441918105,
                "max": 7.009602441918105
              },
              "first_final_s": {
                "n": 1,
                "min": 7.140266132890247,
                "median": 7.140266132890247,
                "max": 7.140266132890247
              },
              "completion_s": {
                "n": 1,
                "min": 8.561044022906572,
                "median": 8.561044022906572,
                "max": 8.561044022906572
              }
            },
            "resources": {
              "sample_count": 0,
              "sampled_gpu_peak_mib_by_index": {},
              "minimum_host_memavailable_bytes": null,
              "qualification": "10-second samples during the condition; not an exact peak. Host available RAM is host-wide, not model allocation."
            },
            "rows": [
              {
                "request_id": "glm-capacity-131072-c1-0000",
                "case_id": "capacity-glm-131072-00",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130513,
                "actual_prompt_tokens": 130513,
                "actual_completion_tokens": 177,
                "actual_reasoning_tokens": 17,
                "actual_total_tokens": 130690,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 7.009602441918105,
                "first_final_s": 7.140266132890247,
                "completion_s": 8.561044022906572
              }
            ],
            "effective_timeout_s": {
              "n": 1,
              "median": 1799.700049513951,
              "p95_nearest_rank": 1799.700049513951,
              "min": 1799.700049513951,
              "max": 1799.700049513951
            },
            "deadline_censored": 0
          },
          {
            "condition": "glm-capacity-131072-c10",
            "concurrency": 10,
            "planned": 10,
            "returned": 10,
            "condition_closed": true,
            "not_returned": 0,
            "complete": 10,
            "numeric_citation_pass": 10,
            "statuses": {
              "complete": 10
            },
            "noncompletion_fraction_of_returned": 0.0,
            "completion_fraction_of_planned": 1.0,
            "successful_latency_s": {
              "first_generated_s": {
                "n": 10,
                "min": 7.047012215014547,
                "median": 38.08290115999989,
                "max": 68.96101976605132
              },
              "first_final_s": {
                "n": 10,
                "min": 14.982183127081953,
                "median": 43.15861298603704,
                "max": 69.39771528402343
              },
              "completion_s": {
                "n": 10,
                "min": 49.70528297999408,
                "median": 69.76216853997903,
                "max": 71.82875734602567
              }
            },
            "resources": {
              "sample_count": 0,
              "sampled_gpu_peak_mib_by_index": {},
              "minimum_host_memavailable_bytes": null,
              "qualification": "10-second samples during the condition; not an exact peak. Host available RAM is host-wide, not model allocation."
            },
            "rows": [
              {
                "request_id": "glm-capacity-131072-c10-0000",
                "case_id": "capacity-glm-131072-00",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130513,
                "actual_prompt_tokens": 130513,
                "actual_completion_tokens": 168,
                "actual_reasoning_tokens": 19,
                "actual_total_tokens": 130681,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 13.92695149895735,
                "first_final_s": 17.901357238995843,
                "completion_s": 49.70528297999408
              },
              {
                "request_id": "glm-capacity-131072-c10-0006",
                "case_id": "capacity-glm-131072-06",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130513,
                "actual_prompt_tokens": 130513,
                "actual_completion_tokens": 245,
                "actual_reasoning_tokens": 37,
                "actual_total_tokens": 130758,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 7.047012215014547,
                "first_final_s": 14.982183127081953,
                "completion_s": 59.336868235026486
              },
              {
                "request_id": "glm-capacity-131072-c10-0001",
                "case_id": "capacity-glm-131072-01",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130513,
                "actual_prompt_tokens": 130513,
                "actual_completion_tokens": 168,
                "actual_reasoning_tokens": 17,
                "actual_total_tokens": 130681,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 27.745185902924277,
                "first_final_s": 31.28905204392504,
                "completion_s": 63.50312904594466
              },
              {
                "request_id": "glm-capacity-131072-c10-0007",
                "case_id": "capacity-glm-131072-07",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130513,
                "actual_prompt_tokens": 130513,
                "actual_completion_tokens": 204,
                "actual_reasoning_tokens": 37,
                "actual_total_tokens": 130717,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 20.822771701030433,
                "first_final_s": 28.807336093042977,
                "completion_s": 64.29743270901963
              },
              {
                "request_id": "glm-capacity-131072-c10-0003",
                "case_id": "capacity-glm-131072-03",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130513,
                "actual_prompt_tokens": 130513,
                "actual_completion_tokens": 193,
                "actual_reasoning_tokens": 20,
                "actual_total_tokens": 130706,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 34.63869878393598,
                "first_final_s": 38.85620647401083,
                "completion_s": 69.34710641403217
              },
              {
                "request_id": "glm-capacity-131072-c10-0004",
                "case_id": "capacity-glm-131072-04",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130513,
                "actual_prompt_tokens": 130513,
                "actual_completion_tokens": 167,
                "actual_reasoning_tokens": 20,
                "actual_total_tokens": 130680,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 55.327338491915725,
                "first_final_s": 59.567827166989446,
                "completion_s": 70.17723066592589
              },
              {
                "request_id": "glm-capacity-131072-c10-0008",
                "case_id": "capacity-glm-131072-08",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130513,
                "actual_prompt_tokens": 130513,
                "actual_completion_tokens": 234,
                "actual_reasoning_tokens": 27,
                "actual_total_tokens": 130747,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 41.527103536063805,
                "first_final_s": 47.46101949806325,
                "completion_s": 70.20433478208724
              },
              {
                "request_id": "glm-capacity-131072-c10-0005",
                "case_id": "capacity-glm-131072-05",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130513,
                "actual_prompt_tokens": 130513,
                "actual_completion_tokens": 217,
                "actual_reasoning_tokens": 36,
                "actual_total_tokens": 130730,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 48.43034447391983,
                "first_final_s": 56.213511933921836,
                "completion_s": 70.3659705669852
              },
              {
                "request_id": "glm-capacity-131072-c10-0009",
                "case_id": "capacity-glm-131072-09",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130513,
                "actual_prompt_tokens": 130513,
                "actual_completion_tokens": 235,
                "actual_reasoning_tokens": 37,
                "actual_total_tokens": 130748,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 68.96101976605132,
                "first_final_s": 69.39771528402343,
                "completion_s": 71.44866454799194
              },
              {
                "request_id": "glm-capacity-131072-c10-0002",
                "case_id": "capacity-glm-131072-02",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130513,
                "actual_prompt_tokens": 130513,
                "actual_completion_tokens": 307,
                "actual_reasoning_tokens": 37,
                "actual_total_tokens": 130820,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 62.2316159439506,
                "first_final_s": 69.03022235492244,
                "completion_s": 71.82875734602567
              }
            ],
            "effective_timeout_s": {
              "n": 10,
              "median": 1562.8773140130797,
              "p95_nearest_rank": 1562.8839032721007,
              "min": 1562.8709845690755,
              "max": 1562.8839032721007
            },
            "deadline_censored": 0
          },
          {
            "condition": "glm-capacity-131072-c20",
            "concurrency": 20,
            "planned": 20,
            "returned": 20,
            "condition_closed": true,
            "not_returned": 0,
            "complete": 20,
            "numeric_citation_pass": 19,
            "statuses": {
              "complete": 20
            },
            "noncompletion_fraction_of_returned": 0.0,
            "completion_fraction_of_planned": 1.0,
            "successful_latency_s": {
              "first_generated_s": {
                "n": 20,
                "min": 7.053054201998748,
                "median": 72.41582305950578,
                "max": 138.26176876807585
              },
              "first_final_s": {
                "n": 20,
                "min": 11.226864618016407,
                "median": 79.3432776874979,
                "max": 138.46964352508076
              },
              "completion_s": {
                "n": 20,
                "min": 51.121960510034114,
                "median": 118.44831668649567,
                "max": 140.31008140207268
              }
            },
            "resources": {
              "sample_count": 0,
              "sampled_gpu_peak_mib_by_index": {},
              "minimum_host_memavailable_bytes": null,
              "qualification": "10-second samples during the condition; not an exact peak. Host available RAM is host-wide, not model allocation."
            },
            "rows": [
              {
                "request_id": "glm-capacity-131072-c20-0003",
                "case_id": "capacity-glm-131072-03",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130513,
                "actual_prompt_tokens": 130513,
                "actual_completion_tokens": 207,
                "actual_reasoning_tokens": 20,
                "actual_total_tokens": 130720,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 7.053054201998748,
                "first_final_s": 11.226864618016407,
                "completion_s": 51.121960510034114
              },
              {
                "request_id": "glm-capacity-131072-c20-0015",
                "case_id": "capacity-glm-131072-15",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130513,
                "actual_prompt_tokens": 130513,
                "actual_completion_tokens": 198,
                "actual_reasoning_tokens": 18,
                "actual_total_tokens": 130711,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 13.916219739010558,
                "first_final_s": 17.656960383988917,
                "completion_s": 56.21696943801362
              },
              {
                "request_id": "glm-capacity-131072-c20-0017",
                "case_id": "capacity-glm-131072-17",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130513,
                "actual_prompt_tokens": 130513,
                "actual_completion_tokens": 208,
                "actual_reasoning_tokens": 29,
                "actual_total_tokens": 130721,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 20.787501474958844,
                "first_final_s": 27.19473896897398,
                "completion_s": 65.12248686992098
              },
              {
                "request_id": "glm-capacity-131072-c20-0005",
                "case_id": "capacity-glm-131072-05",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130513,
                "actual_prompt_tokens": 130513,
                "actual_completion_tokens": 166,
                "actual_reasoning_tokens": 22,
                "actual_total_tokens": 130679,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 34.60241481801495,
                "first_final_s": 39.29232592601329,
                "completion_s": 70.03444176597986
              },
              {
                "request_id": "glm-capacity-131072-c20-0002",
                "case_id": "capacity-glm-131072-02",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130513,
                "actual_prompt_tokens": 130513,
                "actual_completion_tokens": 217,
                "actual_reasoning_tokens": 16,
                "actual_total_tokens": 130730,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 27.691681722993962,
                "first_final_s": 31.01767344889231,
                "completion_s": 74.11967329494655
              },
              {
                "request_id": "glm-capacity-131072-c20-0010",
                "case_id": "capacity-glm-131072-10",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130513,
                "actual_prompt_tokens": 130513,
                "actual_completion_tokens": 175,
                "actual_reasoning_tokens": 17,
                "actual_total_tokens": 130688,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 41.497022668947466,
                "first_final_s": 45.05456136201974,
                "completion_s": 78.73557488096412
              },
              {
                "request_id": "glm-capacity-131072-c20-0012",
                "case_id": "capacity-glm-131072-12",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130513,
                "actual_prompt_tokens": 130513,
                "actual_completion_tokens": 184,
                "actual_reasoning_tokens": 20,
                "actual_total_tokens": 130697,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 48.412209950969554,
                "first_final_s": 52.67079499794636,
                "completion_s": 87.67657478095498
              },
              {
                "request_id": "glm-capacity-131072-c20-0001",
                "case_id": "capacity-glm-131072-01",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130513,
                "actual_prompt_tokens": 130513,
                "actual_completion_tokens": 180,
                "actual_reasoning_tokens": 17,
                "actual_total_tokens": 130693,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 55.348003681981936,
                "first_final_s": 58.911531200050376,
                "completion_s": 93.67841060797218
              },
              {
                "request_id": "glm-capacity-131072-c20-0008",
                "case_id": "capacity-glm-131072-08",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130513,
                "actual_prompt_tokens": 130513,
                "actual_completion_tokens": 248,
                "actual_reasoning_tokens": 27,
                "actual_total_tokens": 130761,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 62.22961717902217,
                "first_final_s": 68.19413938594516,
                "completion_s": 110.60626466094982
              },
              {
                "request_id": "glm-capacity-131072-c20-0000",
                "case_id": "capacity-glm-131072-00",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130513,
                "actual_prompt_tokens": 130513,
                "actual_completion_tokens": 181,
                "actual_reasoning_tokens": 17,
                "actual_total_tokens": 130694,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 82.78984713798855,
                "first_final_s": 86.31228325795382,
                "completion_s": 116.28028121602256
              },
              {
                "request_id": "glm-capacity-131072-c20-0009",
                "case_id": "capacity-glm-131072-09",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130513,
                "actual_prompt_tokens": 130513,
                "actual_completion_tokens": 266,
                "actual_reasoning_tokens": 37,
                "actual_total_tokens": 130779,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 68.97013486200012,
                "first_final_s": 76.9308236859506,
                "completion_s": 120.61635215696879
              },
              {
                "request_id": "glm-capacity-131072-c20-0016",
                "case_id": "capacity-glm-131072-16",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130513,
                "actual_prompt_tokens": 130513,
                "actual_completion_tokens": 267,
                "actual_reasoning_tokens": 27,
                "actual_total_tokens": 130780,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 75.86151125701144,
                "first_final_s": 81.7557316890452,
                "completion_s": 127.75194690097123
              },
              {
                "request_id": "glm-capacity-131072-c20-0007",
                "case_id": "capacity-glm-131072-07",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130513,
                "actual_prompt_tokens": 130513,
                "actual_completion_tokens": 181,
                "actual_reasoning_tokens": 19,
                "actual_total_tokens": 130694,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 96.59119455807377,
                "first_final_s": 100.59170884196647,
                "completion_s": 130.11795796907973
              },
              {
                "request_id": "glm-capacity-131072-c20-0011",
                "case_id": "capacity-glm-131072-11",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130513,
                "actual_prompt_tokens": 130513,
                "actual_completion_tokens": 217,
                "actual_reasoning_tokens": 45,
                "actual_total_tokens": 130730,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 89.67056038498413,
                "first_final_s": 99.25984703598078,
                "completion_s": 131.1170601330232
              },
              {
                "request_id": "glm-capacity-131072-c20-0013",
                "case_id": "capacity-glm-131072-13",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130513,
                "actual_prompt_tokens": 130513,
                "actual_completion_tokens": 208,
                "actual_reasoning_tokens": 28,
                "actual_total_tokens": 130721,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 103.50187944306526,
                "first_final_s": 109.6637261849828,
                "completion_s": 138.4933652060572
              },
              {
                "request_id": "glm-capacity-131072-c20-0019",
                "case_id": "capacity-glm-131072-19",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130513,
                "actual_prompt_tokens": 130513,
                "actual_completion_tokens": 189,
                "actual_reasoning_tokens": 19,
                "actual_total_tokens": 130702,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 110.31212600192521,
                "first_final_s": 110.5355999879539,
                "completion_s": 138.63977637689095
              },
              {
                "request_id": "glm-capacity-131072-c20-0006",
                "case_id": "capacity-glm-131072-06",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130513,
                "actual_prompt_tokens": 130513,
                "actual_completion_tokens": 209,
                "actual_reasoning_tokens": 25,
                "actual_total_tokens": 130722,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 117.53890951001085,
                "first_final_s": 122.93638854799792,
                "completion_s": 139.4635765150888
              },
              {
                "request_id": "glm-capacity-131072-c20-0014",
                "case_id": "capacity-glm-131072-14",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130513,
                "actual_prompt_tokens": 130513,
                "actual_completion_tokens": 214,
                "actual_reasoning_tokens": 0,
                "actual_total_tokens": 130727,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": false,
                "first_generated_s": 124.58803398895543,
                "first_final_s": 124.58803398895543,
                "completion_s": 139.83903639600612
              },
              {
                "request_id": "glm-capacity-131072-c20-0018",
                "case_id": "capacity-glm-131072-18",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130513,
                "actual_prompt_tokens": 130513,
                "actual_completion_tokens": 190,
                "actual_reasoning_tokens": 18,
                "actual_total_tokens": 130703,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 131.35992623190396,
                "first_final_s": 135.1159454738954,
                "completion_s": 139.91120947396848
              },
              {
                "request_id": "glm-capacity-131072-c20-0004",
                "case_id": "capacity-glm-131072-04",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130513,
                "actual_prompt_tokens": 130513,
                "actual_completion_tokens": 200,
                "actual_reasoning_tokens": 18,
                "actual_total_tokens": 130713,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 138.26176876807585,
                "first_final_s": 138.46964352508076,
                "completion_s": 140.31008140207268
              }
            ],
            "effective_timeout_s": {
              "n": 20,
              "median": 1491.0339536310057,
              "p95_nearest_rank": 1491.047125459998,
              "min": 1491.020331340027,
              "max": 1491.0487557119923
            },
            "deadline_censored": 0
          },
          {
            "condition": "glm-capacity-131072-c5",
            "concurrency": 5,
            "planned": 5,
            "returned": 5,
            "condition_closed": true,
            "not_returned": 0,
            "complete": 5,
            "numeric_citation_pass": 5,
            "statuses": {
              "complete": 5
            },
            "noncompletion_fraction_of_returned": 0.0,
            "completion_fraction_of_planned": 1.0,
            "successful_latency_s": {
              "first_generated_s": {
                "n": 5,
                "min": 7.004863991052844,
                "median": 20.751437302911654,
                "max": 34.47916573099792
              },
              "first_final_s": {
                "n": 5,
                "min": 10.497842377983034,
                "median": 24.282761360984296,
                "max": 34.87998083396815
              },
              "completion_s": {
                "n": 5,
                "min": 34.93017001100816,
                "median": 35.65816846303642,
                "max": 36.793262933031656
              }
            },
            "resources": {
              "sample_count": 0,
              "sampled_gpu_peak_mib_by_index": {},
              "minimum_host_memavailable_bytes": null,
              "qualification": "10-second samples during the condition; not an exact peak. Host available RAM is host-wide, not model allocation."
            },
            "rows": [
              {
                "request_id": "glm-capacity-131072-c5-0001",
                "case_id": "capacity-glm-131072-01",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130513,
                "actual_prompt_tokens": 130513,
                "actual_completion_tokens": 169,
                "actual_reasoning_tokens": 17,
                "actual_total_tokens": 130682,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 7.004863991052844,
                "first_final_s": 10.497842377983034,
                "completion_s": 34.93017001100816
              },
              {
                "request_id": "glm-capacity-131072-c5-0000",
                "case_id": "capacity-glm-131072-00",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130513,
                "actual_prompt_tokens": 130513,
                "actual_completion_tokens": 160,
                "actual_reasoning_tokens": 17,
                "actual_total_tokens": 130673,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 20.751437302911654,
                "first_final_s": 24.282761360984296,
                "completion_s": 35.50801607291214
              },
              {
                "request_id": "glm-capacity-131072-c5-0003",
                "case_id": "capacity-glm-131072-03",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130513,
                "actual_prompt_tokens": 130513,
                "actual_completion_tokens": 207,
                "actual_reasoning_tokens": 20,
                "actual_total_tokens": 130720,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 13.856553656980395,
                "first_final_s": 18.048477643053047,
                "completion_s": 35.65816846303642
              },
              {
                "request_id": "glm-capacity-131072-c5-0004",
                "case_id": "capacity-glm-131072-04",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130513,
                "actual_prompt_tokens": 130513,
                "actual_completion_tokens": 211,
                "actual_reasoning_tokens": 20,
                "actual_total_tokens": 130724,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 27.638769351062365,
                "first_final_s": 31.865146362106316,
                "completion_s": 36.31231117609423
              },
              {
                "request_id": "glm-capacity-131072-c5-0002",
                "case_id": "capacity-glm-131072-02",
                "target_prompt_tokens": 131072,
                "tokenizer_prompt_tokens": 130513,
                "actual_prompt_tokens": 130513,
                "actual_completion_tokens": 232,
                "actual_reasoning_tokens": 35,
                "actual_total_tokens": 130745,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 34.47916573099792,
                "first_final_s": 34.87998083396815,
                "completion_s": 36.793262933031656
              }
            ],
            "effective_timeout_s": {
              "n": 5,
              "median": 1599.6800362770446,
              "p95_nearest_rank": 1599.6847300691297,
              "min": 1599.6757787290262,
              "max": 1599.6847300691297
            },
            "deadline_censored": 0
          },
          {
            "condition": "glm-capacity-262144-c1",
            "concurrency": 1,
            "planned": 1,
            "returned": 1,
            "condition_closed": true,
            "not_returned": 0,
            "complete": 1,
            "numeric_citation_pass": 1,
            "statuses": {
              "complete": 1
            },
            "noncompletion_fraction_of_returned": 0.0,
            "completion_fraction_of_planned": 1.0,
            "successful_latency_s": {
              "first_generated_s": {
                "n": 1,
                "min": 17.195782989962026,
                "median": 17.195782989962026,
                "max": 17.195782989962026
              },
              "first_final_s": {
                "n": 1,
                "min": 17.298417730955407,
                "median": 17.298417730955407,
                "max": 17.298417730955407
              },
              "completion_s": {
                "n": 1,
                "min": 18.587603739928454,
                "median": 18.587603739928454,
                "max": 18.587603739928454
              }
            },
            "resources": {
              "sample_count": 0,
              "sampled_gpu_peak_mib_by_index": {},
              "minimum_host_memavailable_bytes": null,
              "qualification": "10-second samples during the condition; not an exact peak. Host available RAM is host-wide, not model allocation."
            },
            "rows": [
              {
                "request_id": "glm-capacity-262144-c1-0000",
                "case_id": "capacity-glm-262144-00",
                "target_prompt_tokens": 262144,
                "tokenizer_prompt_tokens": 261503,
                "actual_prompt_tokens": 261503,
                "actual_completion_tokens": 160,
                "actual_reasoning_tokens": 17,
                "actual_total_tokens": 261663,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 17.195782989962026,
                "first_final_s": 17.298417730955407,
                "completion_s": 18.587603739928454
              }
            ],
            "effective_timeout_s": {
              "n": 1,
              "median": 1790.4493065560237,
              "p95_nearest_rank": 1790.4493065560237,
              "min": 1790.4493065560237,
              "max": 1790.4493065560237
            },
            "deadline_censored": 0
          },
          {
            "condition": "glm-capacity-524288-c1",
            "concurrency": 1,
            "planned": 1,
            "returned": 1,
            "condition_closed": true,
            "not_returned": 0,
            "complete": 1,
            "numeric_citation_pass": 1,
            "statuses": {
              "complete": 1
            },
            "noncompletion_fraction_of_returned": 0.0,
            "completion_fraction_of_planned": 1.0,
            "successful_latency_s": {
              "first_generated_s": {
                "n": 1,
                "min": 48.64023239002563,
                "median": 48.64023239002563,
                "max": 48.64023239002563
              },
              "first_final_s": {
                "n": 1,
                "min": 48.86234502703883,
                "median": 48.86234502703883,
                "max": 48.86234502703883
              },
              "completion_s": {
                "n": 1,
                "min": 51.19223574001808,
                "median": 51.19223574001808,
                "max": 51.19223574001808
              }
            },
            "resources": {
              "sample_count": 0,
              "sampled_gpu_peak_mib_by_index": {},
              "minimum_host_memavailable_bytes": null,
              "qualification": "10-second samples during the condition; not an exact peak. Host available RAM is host-wide, not model allocation."
            },
            "rows": [
              {
                "request_id": "glm-capacity-524288-c1-0000",
                "case_id": "capacity-glm-524288-00",
                "target_prompt_tokens": 524288,
                "tokenizer_prompt_tokens": 523851,
                "actual_prompt_tokens": 523851,
                "actual_completion_tokens": 288,
                "actual_reasoning_tokens": 37,
                "actual_total_tokens": 524139,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 48.64023239002563,
                "first_final_s": 48.86234502703883,
                "completion_s": 51.19223574001808
              }
            ],
            "effective_timeout_s": {
              "n": 1,
              "median": 1769.571886471007,
              "p95_nearest_rank": 1769.571886471007,
              "min": 1769.571886471007,
              "max": 1769.571886471007
            },
            "deadline_censored": 0
          },
          {
            "condition": "glm-capacity-786432-c1",
            "concurrency": 1,
            "planned": 1,
            "returned": 1,
            "condition_closed": true,
            "not_returned": 0,
            "complete": 1,
            "numeric_citation_pass": 1,
            "statuses": {
              "complete": 1
            },
            "noncompletion_fraction_of_returned": 0.0,
            "completion_fraction_of_planned": 1.0,
            "successful_latency_s": {
              "first_generated_s": {
                "n": 1,
                "min": 108.28730388509575,
                "median": 108.28730388509575,
                "max": 108.28730388509575
              },
              "first_final_s": {
                "n": 1,
                "min": 108.36581225809641,
                "median": 108.36581225809641,
                "max": 108.36581225809641
              },
              "completion_s": {
                "n": 1,
                "min": 109.85787500301376,
                "median": 109.85787500301376,
                "max": 109.85787500301376
              }
            },
            "resources": {
              "sample_count": 0,
              "sampled_gpu_peak_mib_by_index": {},
              "minimum_host_memavailable_bytes": null,
              "qualification": "10-second samples during the condition; not an exact peak. Host available RAM is host-wide, not model allocation."
            },
            "rows": [
              {
                "request_id": "glm-capacity-786432-c1-0000",
                "case_id": "capacity-glm-786432-00",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785941,
                "actual_prompt_tokens": 785941,
                "actual_completion_tokens": 186,
                "actual_reasoning_tokens": 28,
                "actual_total_tokens": 786127,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 108.28730388509575,
                "first_final_s": 108.36581225809641,
                "completion_s": 109.85787500301376
              }
            ],
            "effective_timeout_s": {
              "n": 1,
              "median": 1714.653564064065,
              "p95_nearest_rank": 1714.653564064065,
              "min": 1714.653564064065,
              "max": 1714.653564064065
            },
            "deadline_censored": 0
          },
          {
            "condition": "glm-capacity-786432-c10",
            "concurrency": 10,
            "planned": 10,
            "returned": 10,
            "condition_closed": true,
            "not_returned": 0,
            "complete": 6,
            "numeric_citation_pass": 6,
            "statuses": {
              "complete": 6,
              "timeout": 4
            },
            "noncompletion_fraction_of_returned": 0.4,
            "completion_fraction_of_planned": 0.6,
            "successful_latency_s": {
              "first_generated_s": {
                "n": 6,
                "min": 108.29694166500121,
                "median": 379.2873609290109,
                "max": 652.1087671529967
              },
              "first_final_s": {
                "n": 6,
                "min": 108.29822990403045,
                "median": 379.31356465793215,
                "max": 652.1166951729683
              },
              "completion_s": {
                "n": 6,
                "min": 109.41624809906352,
                "median": 380.81428697746014,
                "max": 653.680032904027
              }
            },
            "resources": {
              "sample_count": 0,
              "sampled_gpu_peak_mib_by_index": {},
              "minimum_host_memavailable_bytes": null,
              "qualification": "10-second samples during the condition; not an exact peak. Host available RAM is host-wide, not model allocation."
            },
            "rows": [
              {
                "request_id": "glm-capacity-786432-c10-0001",
                "case_id": "capacity-glm-786432-01",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785941,
                "actual_prompt_tokens": 785941,
                "actual_completion_tokens": 138,
                "actual_reasoning_tokens": 17,
                "actual_total_tokens": 786079,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 108.29694166500121,
                "first_final_s": 108.29822990403045,
                "completion_s": 109.41624809906352
              },
              {
                "request_id": "glm-capacity-786432-c10-0004",
                "case_id": "capacity-glm-786432-04",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785941,
                "actual_prompt_tokens": 785941,
                "actual_completion_tokens": 207,
                "actual_reasoning_tokens": 18,
                "actual_total_tokens": 786148,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 216.27885805792175,
                "first_final_s": 216.2800908179488,
                "completion_s": 218.06635869899765
              },
              {
                "request_id": "glm-capacity-786432-c10-0006",
                "case_id": "capacity-glm-786432-06",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785941,
                "actual_prompt_tokens": 785941,
                "actual_completion_tokens": 198,
                "actual_reasoning_tokens": 21,
                "actual_total_tokens": 786139,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 324.99604175798595,
                "first_final_s": 325.01900120591745,
                "completion_s": 326.70804966194555
              },
              {
                "request_id": "glm-capacity-786432-c10-0002",
                "case_id": "capacity-glm-786432-02",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785941,
                "actual_prompt_tokens": 785941,
                "actual_completion_tokens": 158,
                "actual_reasoning_tokens": 21,
                "actual_total_tokens": 786099,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 433.57868010003585,
                "first_final_s": 433.60812810994685,
                "completion_s": 434.92052429297473
              },
              {
                "request_id": "glm-capacity-786432-c10-0007",
                "case_id": "capacity-glm-786432-07",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785941,
                "actual_prompt_tokens": 785941,
                "actual_completion_tokens": 377,
                "actual_reasoning_tokens": 204,
                "actual_total_tokens": 786318,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 541.7680303619709,
                "first_final_s": 543.5551791869802,
                "completion_s": 545.1985403890721
              },
              {
                "request_id": "glm-capacity-786432-c10-0000",
                "case_id": "capacity-glm-786432-00",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785941,
                "actual_prompt_tokens": 785941,
                "actual_completion_tokens": 182,
                "actual_reasoning_tokens": 19,
                "actual_total_tokens": 786123,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 652.1087671529967,
                "first_final_s": 652.1166951729683,
                "completion_s": 653.680032904027
              },
              {
                "request_id": "glm-capacity-786432-c10-0009",
                "case_id": "capacity-glm-786432-09",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785941,
                "actual_prompt_tokens": null,
                "actual_completion_tokens": null,
                "actual_reasoning_tokens": null,
                "actual_total_tokens": null,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": null,
                "status": "timeout",
                "error": null,
                "finish_reason": null,
                "numeric_citation_pass": null,
                "first_generated_s": null,
                "first_final_s": null,
                "completion_s": 737.7663431380643
              },
              {
                "request_id": "glm-capacity-786432-c10-0008",
                "case_id": "capacity-glm-786432-08",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785941,
                "actual_prompt_tokens": null,
                "actual_completion_tokens": null,
                "actual_reasoning_tokens": null,
                "actual_total_tokens": null,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": null,
                "status": "timeout",
                "error": null,
                "finish_reason": null,
                "numeric_citation_pass": null,
                "first_generated_s": null,
                "first_final_s": null,
                "completion_s": 737.7847850410035
              },
              {
                "request_id": "glm-capacity-786432-c10-0005",
                "case_id": "capacity-glm-786432-05",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785941,
                "actual_prompt_tokens": null,
                "actual_completion_tokens": null,
                "actual_reasoning_tokens": null,
                "actual_total_tokens": null,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": null,
                "status": "timeout",
                "error": null,
                "finish_reason": null,
                "numeric_citation_pass": null,
                "first_generated_s": null,
                "first_final_s": null,
                "completion_s": 737.879020648892
              },
              {
                "request_id": "glm-capacity-786432-c10-0003",
                "case_id": "capacity-glm-786432-03",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785941,
                "actual_prompt_tokens": null,
                "actual_completion_tokens": null,
                "actual_reasoning_tokens": null,
                "actual_total_tokens": null,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": null,
                "status": "timeout",
                "error": null,
                "finish_reason": null,
                "numeric_citation_pass": null,
                "first_generated_s": null,
                "first_final_s": null,
                "completion_s": 737.9399111099774
              }
            ],
            "effective_timeout_s": {
              "n": 10,
              "median": 737.778311541013,
              "p95_nearest_rank": 737.8226053620456,
              "min": 737.7359366400633,
              "max": 737.8226053620456
            },
            "deadline_censored": 4
          },
          {
            "condition": "glm-capacity-786432-c5",
            "concurrency": 5,
            "planned": 5,
            "returned": 5,
            "condition_closed": true,
            "not_returned": 0,
            "complete": 5,
            "numeric_citation_pass": 5,
            "statuses": {
              "complete": 5
            },
            "noncompletion_fraction_of_returned": 0.0,
            "completion_fraction_of_planned": 1.0,
            "successful_latency_s": {
              "first_generated_s": {
                "n": 5,
                "min": 108.22409284499008,
                "median": 325.33052581490483,
                "max": 541.9294178260025
              },
              "first_final_s": {
                "n": 5,
                "min": 108.22535799595062,
                "median": 325.4273565309122,
                "max": 541.9564454960637
              },
              "completion_s": {
                "n": 5,
                "min": 110.05513168801554,
                "median": 326.8551504178904,
                "max": 543.4398945390712
              }
            },
            "resources": {
              "sample_count": 0,
              "sampled_gpu_peak_mib_by_index": {},
              "minimum_host_memavailable_bytes": null,
              "qualification": "10-second samples during the condition; not an exact peak. Host available RAM is host-wide, not model allocation."
            },
            "rows": [
              {
                "request_id": "glm-capacity-786432-c5-0001",
                "case_id": "capacity-glm-786432-01",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785941,
                "actual_prompt_tokens": 785941,
                "actual_completion_tokens": 211,
                "actual_reasoning_tokens": 17,
                "actual_total_tokens": 786152,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 108.22409284499008,
                "first_final_s": 108.22535799595062,
                "completion_s": 110.05513168801554
              },
              {
                "request_id": "glm-capacity-786432-c5-0002",
                "case_id": "capacity-glm-786432-02",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785941,
                "actual_prompt_tokens": 785941,
                "actual_completion_tokens": 177,
                "actual_reasoning_tokens": 20,
                "actual_total_tokens": 786118,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 216.9378602100769,
                "first_final_s": 216.94409339409322,
                "completion_s": 218.44691711303312
              },
              {
                "request_id": "glm-capacity-786432-c5-0003",
                "case_id": "capacity-glm-786432-03",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785941,
                "actual_prompt_tokens": 785941,
                "actual_completion_tokens": 178,
                "actual_reasoning_tokens": 29,
                "actual_total_tokens": 786119,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 325.33052581490483,
                "first_final_s": 325.4273565309122,
                "completion_s": 326.8551504178904
              },
              {
                "request_id": "glm-capacity-786432-c5-0000",
                "case_id": "capacity-glm-786432-00",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785941,
                "actual_prompt_tokens": 785941,
                "actual_completion_tokens": 154,
                "actual_reasoning_tokens": 19,
                "actual_total_tokens": 786095,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 433.8198952949606,
                "first_final_s": 433.8211350059137,
                "completion_s": 435.10682175692637
              },
              {
                "request_id": "glm-capacity-786432-c5-0004",
                "case_id": "capacity-glm-786432-04",
                "target_prompt_tokens": 786432,
                "tokenizer_prompt_tokens": 785941,
                "actual_prompt_tokens": 785941,
                "actual_completion_tokens": 178,
                "actual_reasoning_tokens": 21,
                "actual_total_tokens": 786119,
                "requested_output_tokens": 262144,
                "tokenizer_matches_usage": true,
                "status": "complete",
                "error": null,
                "finish_reason": "stop",
                "numeric_citation_pass": true,
                "first_generated_s": 541.9294178260025,
                "first_final_s": 541.9564454960637,
                "completion_s": 543.4398945390712
              }
            ],
            "effective_timeout_s": {
              "n": 5,
              "median": 1281.286044543027,
              "p95_nearest_rank": 1281.3192997170845,
              "min": 1281.2666513461154,
              "max": 1281.3192997170845
            },
            "deadline_censored": 0
          }
        ],
        "largest_complete_input_tokens_c1": 785941,
        "largest_numeric_citation_pass_input_tokens_c1": 785941,
        "largest_all20_complete_input_tokens_lower_bound": 130513,
        "largest_all20_numeric_citation_pass_input_tokens_lower_bound": null,
        "customer_latency_SLO": "Not yet agreed; no practical-limit pass/fail inferred.",
        "unmeasured_prompt_targets": [],
        "status": "CAPACITY_RESULTS_AVAILABLE",
        "unmeasured_conditions": [
          {
            "target_prompt_tokens": 786432,
            "concurrency": 20,
            "reason": "shared1800second capacity-stage budget exhausted; not an observed OOM or hardware limit"
          }
        ]
      },
      "science_grades": [
        {
          "case_id": "reason-05-mixing",
          "request_id": "glm-scientific_reasoning-0004",
          "completion_status": "complete",
          "kind": "numeric",
          "automatic_pass": true,
          "needs_human_review": true,
          "json_format_pass": true,
          "numeric_pass": true,
          "citation_ids_pass": true,
          "parsed_value": 44,
          "expected": 44,
          "human_review_scope": "Units, assumptions, explanation correctness, limitations; automatic numeric pass alone is not full scientific quality."
        },
        {
          "case_id": "reason-01-uncertainty",
          "request_id": "glm-scientific_reasoning-0000",
          "completion_status": "complete",
          "kind": "numeric",
          "automatic_pass": false,
          "needs_human_review": true,
          "json_format_pass": true,
          "numeric_pass": false,
          "citation_ids_pass": true,
          "parsed_value": 0.036,
          "expected": 0.03577708763999663,
          "human_review_scope": "Units, assumptions, explanation correctness, limitations; automatic numeric pass alone is not full scientific quality."
        },
        {
          "case_id": "reason-04-arrhenius",
          "request_id": "glm-scientific_reasoning-0003",
          "completion_status": "complete",
          "kind": "numeric",
          "automatic_pass": true,
          "needs_human_review": true,
          "json_format_pass": true,
          "numeric_pass": true,
          "citation_ids_pass": true,
          "parsed_value": 30.1,
          "expected": 30.141746474355973,
          "human_review_scope": "Units, assumptions, explanation correctness, limitations; automatic numeric pass alone is not full scientific quality."
        },
        {
          "case_id": "reason-06-pump_power",
          "request_id": "glm-scientific_reasoning-0005",
          "completion_status": "complete",
          "kind": "numeric",
          "automatic_pass": true,
          "needs_human_review": true,
          "json_format_pass": true,
          "numeric_pass": true,
          "citation_ids_pass": true,
          "parsed_value": 1000,
          "expected": 1000,
          "human_review_scope": "Units, assumptions, explanation correctness, limitations; automatic numeric pass alone is not full scientific quality."
        },
        {
          "case_id": "reason-02-bayes",
          "request_id": "glm-scientific_reasoning-0001",
          "completion_status": "complete",
          "kind": "numeric",
          "automatic_pass": true,
          "needs_human_review": true,
          "json_format_pass": true,
          "numeric_pass": true,
          "citation_ids_pass": true,
          "parsed_value": 0.161,
          "expected": 0.16101694915254236,
          "human_review_scope": "Units, assumptions, explanation correctness, limitations; automatic numeric pass alone is not full scientific quality."
        },
        {
          "case_id": "reason-03-weighted_mean",
          "request_id": "glm-scientific_reasoning-0002",
          "completion_status": "complete",
          "kind": "numeric",
          "automatic_pass": true,
          "needs_human_review": true,
          "json_format_pass": true,
          "numeric_pass": true,
          "citation_ids_pass": true,
          "parsed_value": 10.8,
          "expected": 10.8,
          "human_review_scope": "Units, assumptions, explanation correctness, limitations; automatic numeric pass alone is not full scientific quality."
        },
        {
          "case_id": "reason-11-unit_conversion",
          "request_id": "glm-scientific_reasoning-0010",
          "completion_status": "complete",
          "kind": "numeric",
          "automatic_pass": true,
          "needs_human_review": true,
          "json_format_pass": true,
          "numeric_pass": true,
          "citation_ids_pass": true,
          "parsed_value": 192,
          "expected": 192,
          "human_review_scope": "Units, assumptions, explanation correctness, limitations; automatic numeric pass alone is not full scientific quality."
        },
        {
          "case_id": "reason-09-diffusion_stability",
          "request_id": "glm-scientific_reasoning-0008",
          "completion_status": "complete",
          "kind": "numeric",
          "automatic_pass": true,
          "needs_human_review": true,
          "json_format_pass": true,
          "numeric_pass": true,
          "citation_ids_pass": true,
          "parsed_value": 0.05,
          "expected": 0.05,
          "human_review_scope": "Units, assumptions, explanation correctness, limitations; automatic numeric pass alone is not full scientific quality."
        },
        {
          "case_id": "reason-07-decay_inverse",
          "request_id": "glm-scientific_reasoning-0006",
          "completion_status": "complete",
          "kind": "numeric",
          "automatic_pass": true,
          "needs_human_review": true,
          "json_format_pass": true,
          "numeric_pass": true,
          "citation_ids_pass": true,
          "parsed_value": 20,
          "expected": 20,
          "human_review_scope": "Units, assumptions, explanation correctness, limitations; automatic numeric pass alone is not full scientific quality."
        },
        {
          "case_id": "reason-08-grid_order",
          "request_id": "glm-scientific_reasoning-0007",
          "completion_status": "complete",
          "kind": "numeric",
          "automatic_pass": true,
          "needs_human_review": true,
          "json_format_pass": true,
          "numeric_pass": true,
          "citation_ids_pass": true,
          "parsed_value": 2,
          "expected": 2,
          "human_review_scope": "Units, assumptions, explanation correctness, limitations; automatic numeric pass alone is not full scientific quality."
        },
        {
          "case_id": "reason-10-correlated_uncertainty",
          "request_id": "glm-scientific_reasoning-0009",
          "completion_status": "complete",
          "kind": "numeric",
          "automatic_pass": false,
          "needs_human_review": true,
          "json_format_pass": true,
          "numeric_pass": false,
          "citation_ids_pass": true,
          "parsed_value": 0.63,
          "expected": 0.6324555320336759,
          "human_review_scope": "Units, assumptions, explanation correctness, limitations; automatic numeric pass alone is not full scientific quality."
        },
        {
          "case_id": "reason-12-serial_yield",
          "request_id": "glm-scientific_reasoning-0011",
          "completion_status": "complete",
          "kind": "numeric",
          "automatic_pass": true,
          "needs_human_review": true,
          "json_format_pass": true,
          "numeric_pass": true,
          "citation_ids_pass": true,
          "parsed_value": 0.6,
          "expected": 0.6,
          "human_review_scope": "Units, assumptions, explanation correctness, limitations; automatic numeric pass alone is not full scientific quality."
        },
        {
          "case_id": "reason-13-confounding",
          "request_id": "glm-scientific_reasoning-0012",
          "completion_status": "complete",
          "kind": "human_rubric",
          "automatic_pass": null,
          "needs_human_review": true,
          "rubric": [
            "材料と温度が交絡し因果効果は未同定",
            "同温度比較または材料×温度の要因実験",
            "無作為化・反復・不確かさ"
          ],
          "json_format_pass": true
        },
        {
          "case_id": "reason-14-missing_heat",
          "request_id": "glm-scientific_reasoning-0013",
          "completion_status": "complete",
          "kind": "human_rubric",
          "automatic_pass": null,
          "needs_human_review": true,
          "rubric": [
            "温度差だけから熱量は決まらない",
            "質量・比熱・容器熱容量が必要",
            "熱損失や外部加熱・反応量と符号の確認"
          ],
          "json_format_pass": true
        },
        {
          "case_id": "reason-16-simpson",
          "request_id": "glm-scientific_reasoning-0015",
          "completion_status": "complete",
          "kind": "human_rubric",
          "automatic_pass": null,
          "needs_human_review": true,
          "rubric": [
            "A全体83% B全体27%",
            "簡単条件は両方90%、困難条件は両方20%",
            "条件構成の違いで全体差が生じる"
          ],
          "json_format_pass": true
        },
        {
          "case_id": "reason-18-overfitting",
          "request_id": "glm-scientific_reasoning-0017",
          "completion_status": "complete",
          "kind": "human_rubric",
          "automatic_pass": null,
          "needs_human_review": true,
          "rubric": [
            "特徴選択と前処理で検証側情報が漏れる",
            "各訓練fold内で前処理と特徴選択をfit",
            "調整にはnested CVまたは独立test"
          ],
          "json_format_pass": true
        },
        {
          "case_id": "reason-20-identifiability",
          "request_id": "glm-scientific_reasoning-0019",
          "completion_status": "complete",
          "kind": "human_rubric",
          "automatic_pass": null,
          "needs_human_review": true,
          "rubric": [
            "識別できるのは積ab=6のみ",
            "(a,b)の複数の組が同じ観測を生成",
            "片方の独立測定または異なる依存性の観測が必要"
          ],
          "json_format_pass": true
        },
        {
          "case_id": "reason-15-pvalue",
          "request_id": "glm-scientific_reasoning-0014",
          "completion_status": "complete",
          "kind": "human_rubric",
          "automatic_pass": null,
          "needs_human_review": true,
          "rubric": [
            "p値は仮説の事後確率ではない",
            "帰無仮説とモデルの下でのデータの極端さ",
            "効果量・区間・多重比較・事前計画が必要"
          ],
          "json_format_pass": true
        },
        {
          "case_id": "reason-17-negative_control",
          "request_id": "glm-scientific_reasoning-0016",
          "completion_status": "complete",
          "kind": "human_rubric",
          "automatic_pass": null,
          "needs_human_review": true,
          "rubric": [
            "温度と撹拌を統制し実測",
            "触媒なし対照および適切な担体対照",
            "独立反復・無作為化・速度の定量と不確かさ"
          ],
          "json_format_pass": true
        },
        {
          "case_id": "reason-19-falsification",
          "request_id": "glm-scientific_reasoning-0018",
          "completion_status": "complete",
          "kind": "human_rubric",
          "automatic_pass": null,
          "needs_human_review": true,
          "rubric": [
            "反証可能な事前予測が不足",
            "比較する代替仮説と測定量・許容範囲",
            "観測後の恣意的な条件変更を避ける"
          ],
          "json_format_pass": true
        }
      ]
    }
  },
  "diagnostic": {
    "verdict": "baseline_loop_not_reproduced_cannot_attribute_improvement_to_effort",
    "arms": {
      "max": {
        "failure_selected": {
          "n": 6,
          "statuses": {
            "complete": 6
          },
          "correct_final": 5,
          "suspected_loops": 0,
          "completion_median_s_completed_only": 458.18220366101013,
          "completion_max_s_completed_only": 865.1547164650401,
          "max_reported_completion_tokens": 69713
        },
        "successful_control": {
          "n": 6,
          "statuses": {
            "complete": 6
          },
          "correct_final": 6,
          "suspected_loops": 0,
          "completion_median_s_completed_only": 215.54842451500008,
          "completion_max_s_completed_only": 376.3422752110055,
          "max_reported_completion_tokens": 28259
        }
      },
      "high": {
        "failure_selected": {
          "n": 6,
          "statuses": {
            "complete": 6
          },
          "correct_final": 6,
          "suspected_loops": 0,
          "completion_median_s_completed_only": 727.3566537994775,
          "completion_max_s_completed_only": 2190.348583554034,
          "max_reported_completion_tokens": 206892
        },
        "successful_control": {
          "n": 6,
          "statuses": {
            "complete": 6
          },
          "correct_final": 6,
          "suspected_loops": 0,
          "completion_median_s_completed_only": 25.46147535246564,
          "completion_max_s_completed_only": 127.20693834300619,
          "max_reported_completion_tokens": 8546
        }
      }
    },
    "pairs": [
      {
        "case_id": "gpqa-035",
        "replicate": 0,
        "seed": 20261060,
        "role": "successful_control",
        "max": {
          "status": "complete",
          "correct": true,
          "parsed_choice": "C",
          "loop_suspected": false,
          "paragraph_repeats": 1,
          "completion_s": 107.23020624206401,
          "usage": {
            "prompt_tokens": 335,
            "total_tokens": 7396,
            "completion_tokens": 7061,
            "completion_tokens_details": {
              "reasoning_tokens": 6772
            }
          }
        },
        "high": {
          "status": "complete",
          "correct": true,
          "parsed_choice": "C",
          "loop_suspected": false,
          "paragraph_repeats": 1,
          "completion_s": 14.05148604104761,
          "usage": {
            "prompt_tokens": 335,
            "total_tokens": 939,
            "completion_tokens": 604,
            "completion_tokens_details": {
              "reasoning_tokens": 170
            }
          }
        }
      },
      {
        "case_id": "gpqa-035",
        "replicate": 1,
        "seed": 20361060,
        "role": "successful_control",
        "max": {
          "status": "complete",
          "correct": true,
          "parsed_choice": "C",
          "loop_suspected": false,
          "paragraph_repeats": 1,
          "completion_s": 160.9547494710423,
          "usage": {
            "prompt_tokens": 335,
            "total_tokens": 11436,
            "completion_tokens": 11101,
            "completion_tokens_details": {
              "reasoning_tokens": 10836
            }
          }
        },
        "high": {
          "status": "complete",
          "correct": true,
          "parsed_choice": "C",
          "loop_suspected": false,
          "paragraph_repeats": 0,
          "completion_s": 12.282040392048657,
          "usage": {
            "prompt_tokens": 335,
            "total_tokens": 832,
            "completion_tokens": 497,
            "completion_tokens_details": {
              "reasoning_tokens": 179
            }
          }
        }
      },
      {
        "case_id": "gpqa-035",
        "replicate": 2,
        "seed": 20461060,
        "role": "successful_control",
        "max": {
          "status": "complete",
          "correct": true,
          "parsed_choice": "C",
          "loop_suspected": false,
          "paragraph_repeats": 1,
          "completion_s": 54.747378319036216,
          "usage": {
            "prompt_tokens": 335,
            "total_tokens": 3683,
            "completion_tokens": 3348,
            "completion_tokens_details": {
              "reasoning_tokens": 3121
            }
          }
        },
        "high": {
          "status": "complete",
          "correct": true,
          "parsed_choice": "C",
          "loop_suspected": false,
          "paragraph_repeats": 0,
          "completion_s": 14.480847776983865,
          "usage": {
            "prompt_tokens": 335,
            "total_tokens": 963,
            "completion_tokens": 628,
            "completion_tokens_details": {
              "reasoning_tokens": 190
            }
          }
        }
      },
      {
        "case_id": "gpqa-036",
        "replicate": 0,
        "seed": 20261061,
        "role": "failure_selected",
        "max": {
          "status": "complete",
          "correct": true,
          "parsed_choice": "C",
          "loop_suspected": false,
          "paragraph_repeats": 1,
          "completion_s": 31.153605459956452,
          "usage": {
            "prompt_tokens": 171,
            "total_tokens": 1885,
            "completion_tokens": 1714,
            "completion_tokens_details": {
              "reasoning_tokens": 1531
            }
          }
        },
        "high": {
          "status": "complete",
          "correct": true,
          "parsed_choice": "C",
          "loop_suspected": false,
          "paragraph_repeats": 1,
          "completion_s": 17.128652925021015,
          "usage": {
            "prompt_tokens": 171,
            "total_tokens": 964,
            "completion_tokens": 793,
            "completion_tokens_details": {
              "reasoning_tokens": 435
            }
          }
        }
      },
      {
        "case_id": "gpqa-036",
        "replicate": 1,
        "seed": 20361061,
        "role": "failure_selected",
        "max": {
          "status": "complete",
          "correct": true,
          "parsed_choice": "C",
          "loop_suspected": false,
          "paragraph_repeats": 1,
          "completion_s": 61.795590730034746,
          "usage": {
            "prompt_tokens": 171,
            "total_tokens": 4008,
            "completion_tokens": 3837,
            "completion_tokens_details": {
              "reasoning_tokens": 3686
            }
          }
        },
        "high": {
          "status": "complete",
          "correct": true,
          "parsed_choice": "C",
          "loop_suspected": false,
          "paragraph_repeats": 1,
          "completion_s": 19.49404612998478,
          "usage": {
            "prompt_tokens": 171,
            "total_tokens": 1113,
            "completion_tokens": 942,
            "completion_tokens_details": {
              "reasoning_tokens": 539
            }
          }
        }
      },
      {
        "case_id": "gpqa-036",
        "replicate": 2,
        "seed": 20461061,
        "role": "failure_selected",
        "max": {
          "status": "complete",
          "correct": true,
          "parsed_choice": "C",
          "loop_suspected": false,
          "paragraph_repeats": 1,
          "completion_s": 80.92967006901745,
          "usage": {
            "prompt_tokens": 171,
            "total_tokens": 5350,
            "completion_tokens": 5179,
            "completion_tokens_details": {
              "reasoning_tokens": 5033
            }
          }
        },
        "high": {
          "status": "complete",
          "correct": true,
          "parsed_choice": "C",
          "loop_suspected": false,
          "paragraph_repeats": 1,
          "completion_s": 17.748228073003702,
          "usage": {
            "prompt_tokens": 171,
            "total_tokens": 999,
            "completion_tokens": 828,
            "completion_tokens_details": {
              "reasoning_tokens": 494
            }
          }
        }
      },
      {
        "case_id": "gpqa-127",
        "replicate": 0,
        "seed": 20261152,
        "role": "successful_control",
        "max": {
          "status": "complete",
          "correct": true,
          "parsed_choice": "C",
          "loop_suspected": false,
          "paragraph_repeats": 1,
          "completion_s": 292.4590943589574,
          "usage": {
            "prompt_tokens": 161,
            "total_tokens": 21431,
            "completion_tokens": 21270,
            "completion_tokens_details": {
              "reasoning_tokens": 21136
            }
          }
        },
        "high": {
          "status": "complete",
          "correct": true,
          "parsed_choice": "C",
          "loop_suspected": false,
          "paragraph_repeats": 1,
          "completion_s": 36.44210292794742,
          "usage": {
            "prompt_tokens": 161,
            "total_tokens": 2228,
            "completion_tokens": 2067,
            "completion_tokens_details": {
              "reasoning_tokens": 1595
            }
          }
        }
      },
      {
        "case_id": "gpqa-127",
        "replicate": 1,
        "seed": 20361152,
        "role": "successful_control",
        "max": {
          "status": "complete",
          "correct": true,
          "parsed_choice": "C",
          "loop_suspected": false,
          "paragraph_repeats": 1,
          "completion_s": 376.3422752110055,
          "usage": {
            "prompt_tokens": 161,
            "total_tokens": 28420,
            "completion_tokens": 28259,
            "completion_tokens_details": {
              "reasoning_tokens": 28119
            }
          }
        },
        "high": {
          "status": "complete",
          "correct": true,
          "parsed_choice": "C",
          "loop_suspected": false,
          "paragraph_repeats": 1,
          "completion_s": 127.20693834300619,
          "usage": {
            "prompt_tokens": 161,
            "total_tokens": 8707,
            "completion_tokens": 8546,
            "completion_tokens_details": {
              "reasoning_tokens": 8389
            }
          }
        }
      },
      {
        "case_id": "gpqa-127",
        "replicate": 2,
        "seed": 20461152,
        "role": "successful_control",
        "max": {
          "status": "complete",
          "correct": true,
          "parsed_choice": "C",
          "loop_suspected": false,
          "paragraph_repeats": 1,
          "completion_s": 270.14209955895785,
          "usage": {
            "prompt_tokens": 161,
            "total_tokens": 20096,
            "completion_tokens": 19935,
            "completion_tokens_details": {
              "reasoning_tokens": 19745
            }
          }
        },
        "high": {
          "status": "complete",
          "correct": true,
          "parsed_choice": "C",
          "loop_suspected": false,
          "paragraph_repeats": 1,
          "completion_s": 90.94908575096633,
          "usage": {
            "prompt_tokens": 161,
            "total_tokens": 6451,
            "completion_tokens": 6290,
            "completion_tokens_details": {
              "reasoning_tokens": 6193
            }
          }
        }
      },
      {
        "case_id": "gpqa-128",
        "replicate": 0,
        "seed": 20261153,
        "role": "failure_selected",
        "max": {
          "status": "complete",
          "correct": true,
          "parsed_choice": "D",
          "loop_suspected": false,
          "paragraph_repeats": 1,
          "completion_s": 835.4347372530028,
          "usage": {
            "prompt_tokens": 2830,
            "total_tokens": 69534,
            "completion_tokens": 66704,
            "completion_tokens_details": {
              "reasoning_tokens": 66594
            }
          }
        },
        "high": {
          "status": "complete",
          "correct": true,
          "parsed_choice": "D",
          "loop_suspected": false,
          "paragraph_repeats": 4,
          "completion_s": 1435.2192614689702,
          "usage": {
            "prompt_tokens": 2830,
            "total_tokens": 126697,
            "completion_tokens": 123867,
            "completion_tokens_details": {
              "reasoning_tokens": 123249
            }
          }
        }
      },
      {
        "case_id": "gpqa-128",
        "replicate": 1,
        "seed": 20361153,
        "role": "failure_selected",
        "max": {
          "status": "complete",
          "correct": false,
          "parsed_choice": "C",
          "loop_suspected": false,
          "paragraph_repeats": 1,
          "completion_s": 865.1547164650401,
          "usage": {
            "prompt_tokens": 2830,
            "total_tokens": 72043,
            "completion_tokens": 69213,
            "completion_tokens_details": {
              "reasoning_tokens": 69093
            }
          }
        },
        "high": {
          "status": "complete",
          "correct": true,
          "parsed_choice": "D",
          "loop_suspected": false,
          "paragraph_repeats": 5,
          "completion_s": 1655.6797810558928,
          "usage": {
            "prompt_tokens": 2830,
            "total_tokens": 149592,
            "completion_tokens": 146762,
            "completion_tokens_details": {
              "reasoning_tokens": 145961
            }
          }
        }
      },
      {
        "case_id": "gpqa-128",
        "replicate": 2,
        "seed": 20461153,
        "role": "failure_selected",
        "max": {
          "status": "complete",
          "correct": true,
          "parsed_choice": "D",
          "loop_suspected": false,
          "paragraph_repeats": 2,
          "completion_s": 861.516140309046,
          "usage": {
            "prompt_tokens": 2830,
            "total_tokens": 72543,
            "completion_tokens": 69713,
            "completion_tokens_details": {
              "reasoning_tokens": 69539
            }
          }
        },
        "high": {
          "status": "complete",
          "correct": true,
          "parsed_choice": "D",
          "loop_suspected": false,
          "paragraph_repeats": 3,
          "completion_s": 2190.348583554034,
          "usage": {
            "prompt_tokens": 2830,
            "total_tokens": 209722,
            "completion_tokens": 206892,
            "completion_tokens_details": {
              "reasoning_tokens": 205995
            }
          }
        }
      }
    ],
    "manual_loop_confirmation_required": false,
    "limits": [
      "4 outcome-selected cases,3 fixed seeds; not a new GPQA score or a population-wide reliability estimate.",
      "Matched settings and seeds do not ensure identical batch schedules or bitwise outputs.",
      "Loop detector covers repeated normalized paragraphs>=200characters appearing>=10times; absence of detection does not exclude semantic repetition.",
      "Completion latency summarizes completed answers only; failures and their duration remain in individual records.",
      "Correct final choice does not validate every statement in reasoning."
    ],
    "input_sha256": "f0b355731e283f1a8c4fa2838c900d0cb2f6fd73c9ecbe8a00b4f9beaf9e4051",
    "plan_sha256": "5864878266226cea06cf60732ed018750c808cc881428c0770f0371cff10c3b0"
  },
  "paired_gpqa_counts": {
    "both_correct": 177,
    "both_incorrect": 9,
    "GLM_unscored": 2,
    "flash_only_correct": 5,
    "glm_only_correct": 5
  },
  "limits": [
    "Single run per frozen problem; no statistical superiority claim.",
    "Original capped results retained; diagnostic retries do not replace GPQA outcomes.",
    "Completed-only latency excludes censored requests; counts and failed durations retained in originals.",
    "Scientific explanation review is Codex-assisted and not blind independent expert grading.",
    "Long-context tests are synthetic retrieval/arithmetic, not evidence of full-document scientific understanding.",
    "GLM 786K C10 is stage-budget censored; C20 unmeasured. No claim of an absolute hardware limit.",
    "Model/tokenizer/attention/KV/effort and memory allocation differ despite shared prompts and limits.",
    "Measured on HGX B200 with 8 GPUs. Results are not guaranteed on other GPUs or configurations."
  ]
}
