Search-SWE
CPU TASK Valid result

Long-document Reranking

GPT-6 Luna · Codex (v0.164.0) · reasoning effort max

Task score (%)33.0
Input Tokens23,608,663including 23,336,756 cached
Output Tokens197,538
Agent time1h 32min
Cost$0.37
Recorded steps144

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
2026-10-24T06:05:42.610298Z
Coding finished
2026-10-24T07:37:47.740340Z
Input (includes cache)
23,608,663
Cached input (subset)
23,336,756
Cache writes (subset)
271,484
Output
197,538
Reasoning (reported separately; not added)
148,178
Recorded responses
141

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": "official_harbor_regrade",
  "task_commit": "1d73c6f9c4a90ba922b7572d9395a6f40661bbb2",
  "verifier_commit": "1d73c6f9c4a90ba922b7572d9395a6f40661bbb2",
  "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-document-reranking-gpt-6-luna-max-4e362465c111
Source trajectory SHA-256
d00ddfccd98d10ff22ea06f79f73426c48dd90e2b218c8ccee8a800e430688c4
Final reward SHA-256
67b2a842d9d0bb19564a9967e9126bcecb731e51b5a43317751949b6e563b365