{
  "reviewed_utc": "2026-11-04T19:00:04.193767+00:00",
  "eligible_work_remaining": 0,
  "model_responses": 769,
  "planned_slots": 960,
  "missing_glm_slots": 191,
  "attempts": 803,
  "additional_attempts": 33,
  "attempt_count_distribution": {
    "4": 2,
    "3": 1,
    "1": 742,
    "2": 25
  },
  "four_attempt_exceptions": [
    {
      "model": "z-ai/glm-5.3",
      "id": "groundlm-calibration-v1-exact_walk_count-n32-r100-original__tokens16384",
      "attempts": 4
    },
    {
      "model": "z-ai/glm-5.3",
      "id": "groundlm-calibration-v1-exact_walk_count-n32-r100-original__tokens32768",
      "attempts": 4
    }
  ],
  "cost_by_source_usd": {
    "provider": 2.061980766945,
    "uncertain_upper_bound": 0.12875653,
    "pre_generation_rejection": 0.0
  },
  "accounted_total_usd": 2.190737296945,
  "outstanding_reservations_usd": 0,
  "exact_certificates_recomputed": 24,
  "frozen_hashes_verified": true,
  "summary_and_transitions_verified": true,
  "archive_crc_and_required_sources_verified": true,
  "usage_totals": {
    "prompt_tokens": 1229415,
    "completion_tokens": 11150619,
    "total_tokens": 12380034
  },
  "chart_scope": "Four configurations with 192/192 responses; GLM omitted, not assigned zero",
  "limitations": [
    "24 development graphs; four correlated conditions per graph. Do not rank models using this calibration.",
    "Original accuracy intervals use 24 graph questions; agreement and survival intervals show their eligible graph counts.",
    "All-condition rates and budget transitions are descriptive; their 96 prompt slots are not independent graphs.",
    "A correct final answer counts as success even if the provider also reports a length finish reason.",
    "Completed-pair diagnostics exclude unavailable pairs and are secondary to full-coverage outcomes.",
    "The shared token request policy does not establish equal native compute. No future scored budget is selected automatically."
  ]
}
