Skip to content

Leaderboard Submission: GPT-5.6 Terra (max) + Codex#115

Merged
alexgshaw merged 3 commits into
mainfrom
submission/pr-114
Jul 12, 2026
Merged

Leaderboard Submission: GPT-5.6 Terra (max) + Codex#115
alexgshaw merged 3 commits into
mainfrom
submission/pr-114

Conversation

@github-actions

Copy link
Copy Markdown
Contributor

Promoted from #114 (author @alexgshaw, head 16149aa).
Trials are cloned into leaderboard-owned copies; the checks below ran against the cloned trials.

Maintainer comments /judge then /apply to recompute metrics with disqualified trajectories.


Static Analysis [PASS]

9 checks passed
Check Result
Valid source_filter
Valid metadata
timeout_multiplier default (None or 1.0)
No agent timeout overrides
No verifier timeout overrides
No resource overrides
Valid trial count (89 tasks × ≥5)
Valid per-trial records
Valid task digests

Trial Summary

Result Count
No error 433
AgentTimeoutError 12
Total 445

Submission Summary

Model [org] Effort Agent [org] Date Accuracy ± SE Tokens Cost
GPT-5.6 Terra [OpenAI] max Codex [OpenAI] 2026-07-11 78.65% 1.23% 902,641,810 $421.15

Ran on 16149aa. Errored trials count reward 0 in the metric computation.

Co-authored-by: alexgshaw <47223609+alexgshaw@users.noreply.github.com>
@github-actions github-actions Bot added the lb-submission Leaderboard submission PR label Jul 12, 2026
@alexgshaw

Copy link
Copy Markdown
Contributor

/judge

@github-actions

github-actions Bot commented Jul 12, 2026

Copy link
Copy Markdown
Contributor Author

Harness cheating 🔴 [maintainer reviews - renders a submission invalid]

Reward hacking 🟡 [submitter comments on any false positive - maintainer can uncheck and /apply]

Refusals ⚪ [for information - sometimes a cause of agent failure]

  • none

No Trajectory ⚫ [infra error - no trajectory to judge]

  • none
Ran on 0a2a493. 350 trajectories judged, 3 flagged. Judge job: Job. ✅ 1 disqualified trials applied at 51833ae.
@alexgshaw

Copy link
Copy Markdown
Contributor

/apply

@github-actions

Copy link
Copy Markdown
Contributor Author

Applied Disqualifications on 1 trial(s)

Accuracy 78.65% (± 1.23%) → 78.43% (± 1.25%)

Submission Summary

Model [org] Effort Agent [org] Date Accuracy ± SE Tokens Cost
GPT-5.6 Terra [OpenAI] max Codex [OpenAI] 2026-07-11 78.43% 1.25% 902,641,810 $421.15

Ran on 51833ae. Submission json updated with disqualified_trials and updated metrics.

@alexgshaw
alexgshaw merged commit d49e28f into main Jul 12, 2026
@github-actions
github-actions Bot deleted the submission/pr-114 branch July 12, 2026 03:57
@github-actions

Copy link
Copy Markdown
Contributor Author

✅ Entry submitted to the leaderboard

Model [org] Effort Agent [org] Accuracy Tokens Cost PR Hacks
GPT-5.6 Terra [OpenAI] max Codex [OpenAI] 78.4% ± 1.3% 902,641,810 $421.15 #115 -0.2%

Ran on d49e28f.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

lb-submission Leaderboard submission PR

1 participant