{
  "model_details": {
    "provider": "requesty",
    "model_requested": "gpt-5.4-pro",
    "model_for_requests": "gpt-5.4-pro",
    "api_base_url": "https://router.requesty.ai/v1",
    "chat_completions_endpoint": "https://router.requesty.ai/v1/chat/completions"
  },
  "cache_padding": {
    "enabled": false,
    "target_shared_prefix_tokens": 0,
    "calibration_shared_prefix_tokens": null,
    "target_prompt_tokens": 0,
    "calibration_prompt_tokens": null,
    "calibration_example_id": null,
    "applied_padding_tokens_estimate": 0,
    "examples_with_padding_applied": 0
  },
  "request_control_summary": {
    "configured": {},
    "attempts_total": 1203,
    "attempts_with_control_telemetry": 1200,
    "per_control": {
      "reasoning_effort": {
        "configured_value": null,
        "requested_attempts": 0,
        "sent_attempts": 0,
        "accepted_attempts": 0,
        "rejected_attempts": 0,
        "missing_from_final_request_attempts": 0,
        "acceptance_rate": null,
        "rejected_reasons": {},
        "rejected_example_ids": []
      },
      "thinking_level": {
        "configured_value": null,
        "requested_attempts": 0,
        "sent_attempts": 0,
        "accepted_attempts": 0,
        "rejected_attempts": 0,
        "missing_from_final_request_attempts": 0,
        "acceptance_rate": null,
        "rejected_reasons": {},
        "rejected_example_ids": []
      },
      "effort": {
        "configured_value": null,
        "requested_attempts": 0,
        "sent_attempts": 0,
        "accepted_attempts": 0,
        "rejected_attempts": 0,
        "missing_from_final_request_attempts": 0,
        "acceptance_rate": null,
        "rejected_reasons": {},
        "rejected_example_ids": []
      },
      "verbosity": {
        "configured_value": null,
        "requested_attempts": 0,
        "sent_attempts": 0,
        "accepted_attempts": 0,
        "rejected_attempts": 0,
        "missing_from_final_request_attempts": 0,
        "acceptance_rate": null,
        "rejected_reasons": {},
        "rejected_example_ids": []
      },
      "prompt_cache_key": {
        "configured_value": null,
        "requested_attempts": 0,
        "sent_attempts": 0,
        "accepted_attempts": 0,
        "rejected_attempts": 0,
        "missing_from_final_request_attempts": 0,
        "acceptance_rate": null,
        "rejected_reasons": {},
        "rejected_example_ids": []
      },
      "gemini_cached_content": {
        "configured_value": null,
        "requested_attempts": 0,
        "sent_attempts": 0,
        "accepted_attempts": 0,
        "rejected_attempts": 0,
        "missing_from_final_request_attempts": 0,
        "acceptance_rate": null,
        "rejected_reasons": {},
        "rejected_example_ids": []
      },
      "requesty_auto_cache": {
        "configured_value": null,
        "requested_attempts": 0,
        "sent_attempts": 0,
        "accepted_attempts": 0,
        "rejected_attempts": 0,
        "missing_from_final_request_attempts": 0,
        "acceptance_rate": null,
        "rejected_reasons": {},
        "rejected_example_ids": []
      }
    }
  },
  "usage_metadata_summary": {
    "attempts_total": 1203,
    "attempts_with_usage_metadata": 1200,
    "attempts_with_cached_token_signals": 1200,
    "cached_tokens_total_estimate": 0,
    "cache_read_tokens_total": 0,
    "cache_write_tokens_total": 0,
    "cache_token_fields_totals": {
      "usage.prompt_tokens_details.cached_tokens": 0
    },
    "attempts_with_gemini_cached_content_token_signals": 0,
    "gemini_cached_content_token_count_total": 0,
    "gemini_cached_content_token_fields_totals": {}
  },
  "token_usage_totals": {
    "attempts_total": 1203,
    "attempts_with_token_usage": 1199,
    "attempts_with_output_tokens": 1199,
    "attempts_with_cached_input_tokens": 0,
    "attempts_with_thinking_tokens": 0,
    "input_tokens_total": 844891,
    "cached_input_tokens_total": 0,
    "non_cached_input_tokens_total": 844891,
    "output_tokens_total": 141712,
    "thinking_tokens_total": 0,
    "output_tokens_definition": "total_tokens - prompt_tokens (or completion_tokens + thinking_tokens fallback)"
  },
  "truth_label_count": 1200,
  "prediction_count": 1200,
  "evaluated_example_count": 1200,
  "calibration_metrics": {
    "available": true,
    "sample_count": 1199,
    "bin_count": 10,
    "ece": 0.02292743953295685,
    "mce": 0.4000000000000001,
    "brier_score": 0.04491718098415346
  },
  "first_prompt_timestamp": "2026-03-10T13:52:04.694015Z",
  "last_prompt_timestamp": "2026-03-10T23:14:56.022424Z",
  "overall_time_seconds": 33771.328409,
  "overall_time_human": "9h 22m 51s",
  "accuracy": 0.9491666666666667,
  "macro_precision": 0.6109374164929721,
  "macro_recall": 0.6335729637979327,
  "macro_f1": 0.6211117982702502,
  "per_label": {
    "plural": {
      "precision": 0.8474025974025974,
      "recall": 0.9525547445255474,
      "f1": 0.8969072164948453,
      "support": 274
    },
    "singular": {
      "precision": 0.9854096520763187,
      "recall": 0.9481641468682506,
      "f1": 0.9664281783159053,
      "support": 926
    },
    "unclassified": {
      "precision": 0.0,
      "recall": 0.0,
      "f1": 0.0,
      "support": 0
    }
  },
  "confusion_matrix": {
    "plural": {
      "plural": 261,
      "singular": 13,
      "unclassified": 0
    },
    "singular": {
      "plural": 47,
      "singular": 878,
      "unclassified": 1
    },
    "unclassified": {
      "plural": 0,
      "singular": 0,
      "unclassified": 0
    }
  },
  "labels": [
    "plural",
    "singular",
    "unclassified"
  ],
  "total_examples": 1200,
  "label_metrics_available": true,
  "run_config": {
    "task_name": "OE number",
    "prompt_layout": "standard",
    "task_description": "Determining  grammatical number (sg./pl.) in Old English nouns.",
    "tags": "Old English;morphology;number"
  },
  "source_input_csv": "",
  "source_output_csv": "data/output/OE_number__openai__gpt54__2026-03-10-14-52.csv",
  "source_labels_csv": ""
}