Codex Results by Model
| # | Model | Run | Agent | Mode | Score | Eval | Tests | Est. cost | Calls | Wall | Evidence |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | GPT 5.5 (xhigh) | 20260517T094700Z |
Codex /goal | Mini-SWE-compatible no internet | 79.3% | ok | 341/430 | $4.63 | 72 | 0.25h | manifest · eval json · eval summary · usage audit |
| 2 | GPT 5.5 (xhigh) | 20260518T232006Z |
Codex /goal | Paper prompt + /goal no internet | 32.1% | ok | 138/430 | n/a | 62 | 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 | 32.1% | ok | 138/430 | n/a | 67 | 0.29h | 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
79.3% from 341/430 ProgramBench-scored tests. Raw public eval statuses: failure: 90, passed: 368.
Agent trace summary unavailable.
Likely miss class: behavioral mismatch.
c0ff9908837atests.test_ansimage_api.test_directory_instead_of_file (failure)c0ff9908837atests.test_ansimage_api.test_file_not_found_error (failure)c0ff9908837atests.test_dithering.test_dithering_respects_terminal_size_rows (failure)c0ff9908837atests.test_dithering.test_gradient_chars_dithering_shows_progression (failure)c0ff9908837atests.test_dithering.test_color_image_blocks_dithering_preserves_rgb_values (failure)c0ff9908837atests.test_dithering.test_dithering_mode_2_all_10_brightness_levels (failure)
GPT 5.5 (xhigh) · 20260518T232006Z
32.1% from 138/430 ProgramBench-scored tests. Raw public eval statuses: failure: 4, not_run: 292, passed: 162.
Agent trace: 130 shell command(s), 72 executable-observation command(s), 13 build command(s), 2 blocked attempt(s). Raw Codex logs are not published.
Likely miss class: behavioral mismatch, exact version/output mismatch.
c0ff9908837atests.test_ansimage_api.test_combined_all_flags (not_run)c0ff9908837atests.test_ansimage_api.test_corrupt_png_file_error (not_run)c0ff9908837atests.test_ansimage_api.test_directory_instead_of_file (not_run)c0ff9908837atests.test_ansimage_api.test_dithering_mode_1_minimum_terminal_size (not_run)c0ff9908837atests.test_ansimage_api.test_dithering_mode_2_minimum_terminal_size (not_run)c0ff9908837atests.test_ansimage_api.test_empty_file_error (not_run)
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
| # | Model | Provider | Score | Cost | Calls |
|---|---|---|---|---|---|
| 1 | GPT 5.5 (xhigh) | OpenAI | 85.8% | $7.86 | 63 |
| 2 | Claude Sonnet 4.6 | Anthropic | 74.9% | $24.61 | 335 |
| 3 | GPT 5.5 (high) | OpenAI | 74.2% | $4.15 | 42 |
| 4 | GPT 5.5 | OpenAI | 24.7% | $1.15 | 16 |
| 5 | Gemini 3.1 Pro | 21.6% | $2.32 | 78 | |
| 6 | Claude Opus 4.7 (xhigh) | Anthropic | 20.7% | $30.03 | 224 |
| 7 | Claude Opus 4.6 | Anthropic | 19.5% | $15.01 | 227 |
| 8 | GPT 5.4 | OpenAI | 19.5% | $0.20 | 7 |
| 9 | Claude Opus 4.7 | Anthropic | 19.3% | $5.80 | 92 |
| 10 | GPT 5 mini | OpenAI | 17.9% | $0.02 | 14 |
| 11 | GPT 5.4 mini | OpenAI | 14.4% | $0.03 | 6 |
| 12 | Claude Haiku 4.5 | Anthropic | 9.8% | $0.98 | 136 |
| 13 | Gemini 3 Flash | 8.4% | $0.40 | 83 |