CPU TASK Valid result
Long-PDF Sentence Evidence
GPT-6 Luna · Codex · reasoning effort max
Task score (%)33.3
Input Tokens25,919,795including 25,580,157 cached
Output Tokens273,195
Agent time1h 22min
Cost$0.59
Recorded steps137
Cost is estimated from recorded agent tokens at official API rates in USD (2026-11-01; DeepSeek off-peak pricing).
Loading trajectory…
Select this tab to browse the saved files.
Coding time & token usage
- Original agent completion
- normal
- Coding started
- 2026-10-23T22:59:22.168823Z
- Coding finished
- 2026-10-24T00:21:47.384237Z
- Input (includes cache)
- 25,919,795
- Cached input (subset)
- 25,580,157
- Cache writes (subset)
- 339,239
- Output
- 273,195
- Reasoning (reported separately; not added)
- 206,141
- Recorded responses
- 133
Input includes cached input; output is counted once. Auxiliary reasoning counts are shown as recorded. Regrading does not repeat coding usage.
Select this tab to browse the saved files.
Configuration
Evaluation settings summary
{
"agent": "Codex",
"agent_timeout_sec": null,
"coding_commit": null,
"effort": "max",
"model": "openai/gpt-6-luna",
"provider": "OpenRouter",
"resources": null,
"run_mode": "saved_evidence_uniform_semantic_regrade",
"task_commit": "7e7ce5c052aa4d0cace60dc9a1c419095d521c7c",
"verifier_commit": "7e7ce5c052aa4d0cace60dc9a1c419095d521c7c",
"version": "0.164.0"
}Select this tab to browse the saved files.
Resolved configuration
Select this tab to browse the saved files.
Trial log
Select this tab to browse the saved files.
Exceptions
Select this tab to browse the saved files.
Select this tab to browse the saved files.
Trajectory downloads & provenance
Public text redacts private values and normalizes host paths. Source hashes identify the original records.
- Trial ID
long-pdf-evidence-gpt-6-luna-max-96f6af4d4d99- Source trajectory SHA-256
cd9a3bdccee6355442b0cfdc6c082e6e7466cd0db8126bf835392cc46bad4760- Final reward SHA-256
60cdf1f124bbef1d1f5ab0182f44a2c71c50290915d9caa533ad82cfb0113c60