Codex Results by Model
| # | Model | Run | Agent | Mode | Score | Eval | Tests | Est. cost | Calls | Wall | Evidence |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | GPT 5.5 (xhigh) | 20260519T215046Z |
Codex /goal | Paper prompt + Goal contract no internet | 44.3% | ok | 467/1053 | n/a | 63 | 0.22h | not exported |
| 2 | GPT 5.5 (xhigh) | 20260518T232006Z |
Codex /goal | Paper prompt + /goal no internet | 43.2% | ok | 455/1053 | n/a | 41 | 0.17h | manifest · eval json · eval summary · agent summary · usage audit |
| 3 | GPT 5.5 (xhigh) | 20260517T094700Z |
Codex /goal | Mini-SWE-compatible no internet | 40.3% | ok | 424/1053 | $4.91 | 55 | 0.24h | 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) · 20260519T215046Z
Public eval evidence has not been exported for this row yet.
GPT 5.5 (xhigh) · 20260518T232006Z
43.2% from 455/1053 ProgramBench-scored tests. Raw public eval statuses: failure: 609, passed: 464, skipped: 20.
Agent trace: 41 shell command(s), 20 executable-observation command(s), 2 build command(s), 2 blocked attempt(s). Raw Codex logs are not published.
Likely miss class: behavioral mismatch, exact version/output mismatch.
b0846e00e790tests.test_harvest_closest_extended.test_sortandnaming_t02 (failure)b0846e00e790tests.test_fasta_ops.test_maskfasta_soft_mask (failure)b0846e00e790tests.test_fasta_ops.test_maskfasta_soft_overrides_custom_char (failure)b0846e00e790tests.test_harvest_closest_extended.test_kclosest_t30 (failure)b0846e00e790tests.test_harvest_closest_extended.test_sortandnaming_t19 (failure)b0846e00e790tests.test_harvest_complement_flank_sort_multicov.test_sort_t08 (failure)
manifest · eval json · eval summary · agent summary · usage audit
GPT 5.5 (xhigh) · 20260517T094700Z
40.3% from 424/1053 ProgramBench-scored tests. Raw public eval statuses: failure: 641, passed: 432, skipped: 20.
Agent trace summary unavailable.
Likely miss class: behavioral mismatch, exact version/output mismatch.
b0846e00e790tests.test_fasta_ops.test_maskfasta_soft_mask (failure)b0846e00e790tests.test_fasta_ops.test_maskfasta_soft_overrides_custom_char (failure)b0846e00e790tests.test_harvest_bam_bed_conversions.test_bamtobed_t7_three_blocks_bed12 (failure)b0846e00e790tests.test_harvest_closest_extended.test_sortandnaming_t19 (failure)b0846e00e790tests.test_harvest_closest_extended.test_sortandnaming_t02 (failure)b0846e00e790tests.test_harvest_closest_extended.test_kclosest_t30 (failure)
Official ProgramBench Results by Model
| # | Model | Provider | Score | Cost | Calls |
|---|---|---|---|---|---|
| 1 | GPT 5.5 (xhigh) | OpenAI | 46.1% | $7.07 | 60 |
| 2 | Claude Opus 4.6 | Anthropic | 38.9% | $19.07 | 293 |
| 3 | GPT 5.5 (high) | OpenAI | 38.4% | $4.38 | 41 |
| 4 | GPT 5.5 | OpenAI | 28.8% | $2.08 | 22 |
| 5 | Gemini 3.1 Pro | 20.2% | $1.66 | 73 | |
| 6 | GPT 5.4 | OpenAI | 16.8% | $0.34 | 7 |
| 7 | Gemini 3 Flash | 9.2% | $0.20 | 51 | |
| 8 | Claude Haiku 4.5 | Anthropic | 6.0% | $0.78 | 133 |
| 9 | GPT 5 mini | OpenAI | 4.5% | $0.04 | 17 |
| 10 | GPT 5.4 mini | OpenAI | 4.4% | $0.06 | 9 |
| 11 | Claude Opus 4.7 (xhigh) | Anthropic | 2.1% | $2.85 | 88 |
| 12 | Claude Opus 4.7 | Anthropic | 2.1% | $1.67 | 60 |
| 13 | Claude Sonnet 4.6 | Anthropic | 0.2% | $24.91 | 426 |