ProgramBench task detail

pls-rs__pls.4e1ae50

Task-level results for this Codex /goal scaffold. ProgramBench baseline context is cached from the official task page; Codex scored tests are after active-branch and ignored-test filtering.

332generated behavioral tests
76.5%official best score
69.0%Codex best score
3Codex result rows

Codex Results by Model

#ModelRunAgentModeScoreEvalTestsEst. costCallsWallEvidence
1 GPT 5.5 (xhigh) 20260517T094700Z Codex /goal Mini-SWE-compatible no internet 69.0% ok 229/332 $3.61 75 0.19h manifest · eval json · eval summary · usage audit
2 GPT 5.5 (xhigh) 20260518T232006Z Codex /goal Paper prompt + /goal no internet 67.8% ok 225/332 $4.68 87 0.23h manifest · eval json · eval summary · agent summary · usage audit
3 GPT 5.5 (xhigh) 20260519T215046Z Codex /goal Paper prompt + Goal contract no internet 62.7% ok 208/332 $3.65 58 0.20h not exported

Why Scores Differ

Official ProgramBench rows are the public mini-SWE-agent submissions. GoalBench rows are separate Codex /goal submissions. Same model label does not mean the same agent, prompt, tool loop, or generated implementation. A GoalBench row is resolved only when the ProgramBench-scored pass rate is exactly 100%.

GPT 5.5 (xhigh) · 20260517T094700Z

69.0% from 229/332 ProgramBench-scored tests. Raw public eval statuses: failure: 104, passed: 244, skipped: 6.

Agent trace summary unavailable.

Likely miss class: behavioral mismatch.

  • 4e239660f8fd tests.test_config.test_yaml_with_comments (failure)
  • 4e239660f8fd tests.test_config.test_perm_styles (failure)
  • 4e239660f8fd tests.test_config.test_multiple_icon_choices (failure)
  • 4e239660f8fd tests.test_config.test_empty_config_file (failure)
  • 4e239660f8fd tests.test_config.test_importance_sorting (failure)
  • 4e239660f8fd tests.test_args_output.test_multiple_detail_columns_alignment (failure)

manifest · eval json · eval summary · usage audit

GPT 5.5 (xhigh) · 20260518T232006Z

67.8% from 225/332 ProgramBench-scored tests. Raw public eval statuses: failure: 108, passed: 240, skipped: 6.

Agent trace: 154 shell command(s), 134 executable-observation command(s), 6 build command(s), 2 blocked attempt(s). Raw Codex logs are not published.

Likely miss class: behavioral mismatch.

  • 4e239660f8fd tests.test_config.test_multiple_icon_choices (failure)
  • 4e239660f8fd tests.test_config.test_empty_config_file (failure)
  • 4e239660f8fd tests.test_config.test_perm_styles (failure)
  • 4e239660f8fd tests.test_config.test_yaml_with_comments (failure)
  • 4e239660f8fd tests.test_config.test_importance_sorting (failure)
  • 4e239660f8fd tests.test_args_output.test_detail_user_group_columns (failure)

manifest · eval json · eval summary · agent summary · usage audit

GPT 5.5 (xhigh) · 20260519T215046Z

Public eval evidence has not been exported for this row yet.

Official ProgramBench Results by Model

#ModelProviderScoreCostCalls
1 GPT 5.5 (xhigh) OpenAI 76.5% $10.30 119
2 GPT 5.5 (high) OpenAI 67.2% $3.10 43
3 Gemini 3.1 Pro Google 62.3% $1.60 127
4 Claude Sonnet 4.6 Anthropic 58.4% $40.54 680
5 Claude Opus 4.7 (xhigh) Anthropic 47.3% $7.41 155
6 Claude Opus 4.6 Anthropic 31.0% $23.17 497
7 GPT 5.5 OpenAI 29.8% $0.89 11
8 Claude Opus 4.7 Anthropic 24.1% $1.86 75
9 GPT 5.4 OpenAI 16.9% $0.24 9
10 Claude Haiku 4.5 Anthropic 16.0% $0.77 112
11 Gemini 3 Flash Google 13.9% $0.35 78
12 GPT 5.4 mini OpenAI 9.6% $0.03 9
13 GPT 5 mini OpenAI 2.4% $0.01 7