Codex Results by Model
| # | Model | Run | Agent | Mode | Score | Eval | Tests | Est. cost | Calls | Wall | Evidence |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | GPT 5.5 (xhigh) | 20260518T232006Z |
Codex /goal | Paper prompt + /goal no internet | 66.9% | ok | 938/1402 | n/a | 53 | 0.15h | manifest · eval json · eval summary · agent summary · usage audit |
| 2 | GPT 5.5 (xhigh) | 20260519T215046Z |
Codex /goal | Paper prompt + Goal contract no internet | 62.5% | ok | 876/1402 | n/a | 89 | 0.26h | not exported |
| 3 | GPT 5.5 (xhigh) | 20260517T094700Z |
Codex /goal | Mini-SWE-compatible no internet | 8.6% | warn 1 | 121/1403 | $4.65 | 83 | 0.22h | manifest · eval json · eval 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) · 20260518T232006Z
66.9% from 938/1402 ProgramBench-scored tests. Raw public eval statuses: error: 1, failure: 523, passed: 1303, skipped: 10.
Agent trace: 122 shell command(s), 88 executable-observation command(s), 5 build command(s), 1 blocked attempt(s). Raw Codex logs are not published.
Likely miss class: behavioral mismatch, terminal rendering mismatch.
9514639016aeeval.tests.test_backend_options.test_sqlite_file_path_creates_parent_dir (failure)9514639016aeeval.tests.test_backend_options.test_json_file_path_creates_parent_dir (failure)9514639016aeeval.tests.test_error_paths.test_theme_subcommand_with_no_action (failure)9514639016aeeval.tests.test_tui_comprehensive.test_tui_tag_filter (failure)9514639016aeeval.tests.test_tui_comprehensive.test_tui_backend_type_override (failure)9514639016aeeval.tests.test_tui_comprehensive.test_tui_with_empty_backend (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.
GPT 5.5 (xhigh) · 20260517T094700Z
8.6% from 121/1403 ProgramBench-scored tests. Raw public eval statuses: error: 51, failure: 1482, not_run: 17, passed: 278, skipped: 10.
Agent trace summary unavailable.
Likely miss class: behavioral mismatch, exact version/output mismatch.
9514639016aeeval.tests.test_assign_priority.test_assign_priority_help (failure)9514639016aeeval.tests.test_backend_options.test_sqlite_file_path_creates_parent_dir (failure)9514639016aeeval.tests.test_backend_options.test_backend_type_json_option (failure)9514639016aeeval.tests.test_assign_priority.test_assign_priority_with_positive_number (failure)9514639016aeeval.tests.test_basic_invocation.test_help_short_flag (failure)9514639016aeeval.tests.test_comprehensive_configs.test_export_scroll_per_page_in_config (failure)
Official ProgramBench Results by Model
| # | Model | Provider | Score | Cost | Calls |
|---|---|---|---|---|---|
| 1 | Claude Opus 4.7 (xhigh) | Anthropic | 71.0% | $4.95 | 115 |
| 2 | Claude Opus 4.7 | Anthropic | 70.8% | $4.42 | 107 |
| 3 | GPT 5.5 | OpenAI | 70.0% | $1.26 | 21 |
| 4 | GPT 5.4 | OpenAI | 58.4% | $0.29 | 10 |
| 5 | Claude Haiku 4.5 | Anthropic | 44.2% | $0.65 | 116 |
| 6 | Gemini 3.1 Pro | 41.7% | $0.48 | 42 | |
| 7 | GPT 5 mini | OpenAI | 34.5% | $0.03 | 20 |
| 8 | GPT 5.4 mini | OpenAI | 34.2% | $0.04 | 9 |
| 9 | Claude Sonnet 4.6 | Anthropic | 8.6% | $35.70 | 579 |
| 10 | Gemini 3 Flash | 0.5% | $0.26 | 74 | |
| 11 | Claude Opus 4.6 | Anthropic | 0.0% | $3.27 | 239 |
| 12 | GPT 5.5 (xhigh) | OpenAI | n/a | $6.96 | 63 |
| 13 | GPT 5.5 (high) | OpenAI | n/a | $4.48 | 45 |