1051generated behavioral tests
82.8%official best score
91.8%Codex best score
3Codex result rows
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 | 91.8% | ok | 965/1051 | $7.51 | 121 | 0.29h | manifest · eval json · eval summary · usage audit |
| 2 | GPT 5.5 (xhigh) | 20260518T232006Z |
Codex /goal | Paper prompt + /goal no internet | 86.2% | ok | 906/1051 | n/a | 78 | 0.24h | not exported |
| 3 | GPT 5.5 (xhigh) | 20260519T215046Z |
Codex /goal | Paper prompt + Goal contract no internet | 68.7% | ok | 722/1051 | n/a | 79 | 0.24h | 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
91.8% from 965/1051 ProgramBench-scored tests. Raw public eval statuses: failure: 90, passed: 1285.
Agent trace summary unavailable.
Likely miss class: behavioral mismatch, exact version/output mismatch.
066b68295b09eval.tests.test_completion.test_completion_help (failure)066b68295b09eval.tests.test_completion.test_zsh_completion (failure)066b68295b09eval.tests.test_completion.test_zsh_completion_help (failure)066b68295b09eval.tests.test_info_command.test_info_basic (failure)066b68295b09eval.tests.test_info_command.test_info_with_quiet (failure)066b68295b09eval.tests.test_info_command.test_info_help (failure)
GPT 5.5 (xhigh) · 20260518T232006Z
Public eval evidence has not been exported for this row yet.
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 | Claude Sonnet 4.6 | Anthropic | 82.8% | $15.89 | 416 |
| 2 | GPT 5.5 (xhigh) | OpenAI | 71.1% | $7.05 | 66 |
| 3 | Claude Opus 4.7 (xhigh) | Anthropic | 69.5% | $15.28 | 176 |
| 4 | Claude Opus 4.6 | Anthropic | 69.4% | $10.77 | 253 |
| 5 | GPT 5.5 (high) | OpenAI | 68.0% | $2.92 | 30 |
| 6 | Claude Opus 4.7 | Anthropic | 64.1% | $11.77 | 211 |
| 7 | GPT 5.5 | OpenAI | 62.4% | $1.19 | 17 |
| 8 | Gemini 3 Flash | 53.0% | $0.25 | 62 | |
| 9 | Gemini 3.1 Pro | 40.3% | $1.11 | 70 | |
| 10 | Claude Haiku 4.5 | Anthropic | 31.7% | $0.85 | 132 |
| 11 | GPT 5.4 mini | OpenAI | 28.4% | $0.05 | 24 |
| 12 | GPT 5.4 | OpenAI | 28.3% | $0.66 | 17 |
| 13 | GPT 5 mini | OpenAI | 24.6% | $0.03 | 24 |