ProgramBench task detail

nikoladucak__caps-log.2cf2d1e

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.

551generated behavioral tests
75.3%official best score
55.0%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 55.0% ok 303/551 $7.39 86 0.29h manifest · eval json · eval summary · usage audit
2 GPT 5.5 (xhigh) 20260519T215046Z Codex /goal Paper prompt + Goal contract no internet 46.1% ok 254/551 $8.60 100 0.24h not exported
3 GPT 5.5 (xhigh) 20260518T232006Z Codex /goal Paper prompt + /goal no internet 40.5% ok 223/551 $4.51 68 0.15h manifest · eval json · eval summary · agent summary · usage audit

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

55.0% from 303/551 ProgramBench-scored tests. Raw public eval statuses: failure: 387, passed: 820, skipped: 25.

Agent trace summary unavailable.

Likely miss class: behavioral mismatch.

  • 6bbf9eab4e8f tests.test_comprehensive_behavior.test_encryption_with_binary_content (failure)
  • 6bbf9eab4e8f tests.test_additional_coverage_boost.test_encryption_with_multiple_file_formats (failure)
  • 6bbf9eab4e8f tests.test_comprehensive_behavior.test_multiple_scratchpad_files (failure)
  • 6bbf9eab4e8f tests.test_comprehensive_behavior.test_config_with_log_filename_format (failure)
  • 6bbf9eab4e8f tests.test_additional_coverage.test_encryption_marker_file_content (failure)
  • 6bbf9eab4e8f tests.test_config_error_paths.test_config_invalid_hex_no_hash (failure)

manifest · eval json · eval summary · usage audit

GPT 5.5 (xhigh) · 20260519T215046Z

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

GPT 5.5 (xhigh) · 20260518T232006Z

40.5% from 223/551 ProgramBench-scored tests. Raw public eval statuses: error: 3, failure: 427, not_run: 40, passed: 714, skipped: 25.

Agent trace: 81 shell command(s), 51 executable-observation command(s), 4 build command(s), 3 blocked attempt(s). Raw Codex logs are not published.

Likely miss class: behavioral mismatch.

  • 6bbf9eab4e8f tests.test_basic_invocation.test_invalid_option_format (failure)
  • 6bbf9eab4e8f tests.test_comprehensive_behavior.test_encryption_with_binary_content (failure)
  • 6bbf9eab4e8f tests.test_additional_coverage_boost.test_encryption_with_multiple_file_formats (failure)
  • 6bbf9eab4e8f tests.test_comprehensive_behavior.test_multiple_scratchpad_files (failure)
  • 6bbf9eab4e8f tests.test_comprehensive_behavior.test_config_with_log_filename_format (failure)
  • 6bbf9eab4e8f tests.test_basic_invocation.test_unrecognized_option (failure)

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

Official ProgramBench Results by Model

#ModelProviderScoreCostCalls
1 GPT 5.5 (xhigh) OpenAI 75.3% $8.89 102
2 GPT 5.5 (high) OpenAI 65.5% $5.81 53
3 Gemini 3.1 Pro Google 61.7% $1.70 90
4 Claude Opus 4.6 Anthropic 56.4% $11.12 283
5 GPT 5.5 OpenAI 55.0% $0.96 17
6 Claude Opus 4.7 Anthropic 42.6% $2.40 70
7 GPT 5.4 OpenAI 37.9% $0.29 16
8 Claude Haiku 4.5 Anthropic 34.8% $0.49 97
9 Gemini 3 Flash Google 19.1% $0.36 88
10 GPT 5.4 mini OpenAI 18.9% $0.02 7
11 GPT 5 mini OpenAI 18.9% $0.01 13
12 Claude Opus 4.7 (xhigh) Anthropic 10.0% $4.96 84
13 Claude Sonnet 4.6 Anthropic 10.0% $23.74 425