ProgramBench task detail

incu6us__goimports-reviser.81bd549

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.

513generated behavioral tests
89.7%official best score
81.1%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 81.1% ok 416/513 $4.46 83 0.24h manifest · eval json · eval summary · usage audit
2 GPT 5.5 (xhigh) 20260518T232006Z Codex /goal Paper prompt + /goal no internet 71.0% ok 364/513 n/a 32 0.12h manifest · eval json · eval summary · agent summary · usage audit
3 GPT 5.5 (xhigh) 20260519T215046Z Codex /goal Paper prompt + Goal contract no internet 69.2% ok 355/513 n/a 72 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

81.1% from 416/513 ProgramBench-scored tests. Raw public eval statuses: failure: 103, passed: 491, skipped: 3.

Agent trace summary unavailable.

Likely miss class: behavioral mismatch, exact version/output mismatch.

  • 3b49c4e5d81d eval.tests.test_error_handling.test_malformed_go_file (failure)
  • 3b49c4e5d81d eval.tests.test_import_modifications.test_set_alias_for_versioned_packages (failure)
  • 3b49c4e5d81d eval.tests.test_formatting.test_function_spacing (failure)
  • 3b49c4e5d81d eval.tests.test_input_handling.test_dotdotdot_pattern (failure)
  • 018187d55df6 eval.tests.test_error_conditions.test_nested_directories_with_recursive (failure)
  • 018187d55df6 eval.tests.test_error_conditions.test_excludes_with_absolute_path (failure)

manifest · eval json · eval summary · usage audit

GPT 5.5 (xhigh) · 20260518T232006Z

71.0% from 364/513 ProgramBench-scored tests. Raw public eval statuses: failure: 157, passed: 437, skipped: 3.

Agent trace: 68 shell command(s), 12 executable-observation command(s), 38 build command(s), 1 blocked attempt(s). Raw Codex logs are not published.

Likely miss class: behavioral mismatch, exact version/output mismatch.

  • 3b49c4e5d81d eval.tests.test_error_handling.test_missing_package_declaration (failure)
  • 3b49c4e5d81d eval.tests.test_error_handling.test_malformed_go_file (failure)
  • 3b49c4e5d81d eval.tests.test_formatting.test_format_code_structure (failure)
  • 3b49c4e5d81d eval.tests.test_formatting.test_function_spacing (failure)
  • 3b49c4e5d81d eval.tests.test_import_sorting.test_default_import_order_std_general_project (failure)
  • 3b49c4e5d81d eval.tests.test_input_handling.test_dotdotdot_pattern (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 89.7% $9.03 101
2 Claude Opus 4.7 Anthropic 86.4% $2.60 95
3 Gemini 3 Flash Google 74.1% $0.34 121
4 GPT 5.4 OpenAI 73.3% $0.60 18
5 Claude Sonnet 4.6 Anthropic 72.3% $17.06 441
6 Claude Opus 4.7 (xhigh) Anthropic 71.2% $14.92 266
7 GPT 5.5 (high) OpenAI 69.4% $1.91 21
8 Claude Opus 4.6 Anthropic 68.6% $9.52 230
9 GPT 5.5 OpenAI 59.8% $1.27 24
10 Claude Haiku 4.5 Anthropic 46.6% $1.02 146
11 GPT 5.4 mini OpenAI 14.4% $0.03 11
12 GPT 5 mini OpenAI 8.2% $0.02 12
13 Gemini 3.1 Pro Google 0.0% $1.17 98