906generated behavioral tests
95.4%official best score
92.9%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 | 92.9% | ok | 842/906 | $6.73 | 100 | 0.27h | manifest · eval json · eval summary · usage audit |
| 2 | GPT 5.5 (xhigh) | 20260518T232006Z |
Codex /goal | Paper prompt + /goal no internet | 87.9% | ok | 796/906 | n/a | 66 | 0.22h | not exported |
| 3 | GPT 5.5 (xhigh) | 20260519T215046Z |
Codex /goal | Paper prompt + Goal contract no internet | 46.8% | ok | 424/906 | n/a | 72 | 0.25h | 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
92.9% from 842/906 ProgramBench-scored tests. Raw public eval statuses: failure: 65, passed: 893, skipped: 16.
Agent trace summary unavailable.
Likely miss class: behavioral mismatch, exact version/output mismatch.
43f49b9accc2eval.tests.test_completion.test_completion_zsh (failure)43f49b9accc2eval.tests.test_completion.test_completion_powershell (failure)d2ba62183d6feval.tests.test_basic.test_completion_zsh (failure)850c3224e9b7tests.test_colors_special.test_completion_bash_generates_valid_script (failure)850c3224e9b7tests.test_colors_special.test_completion_output_not_truncated (failure)850c3224e9b7tests.test_display.test_completion_powershell_valid_script (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 | GPT 5.5 (xhigh) | OpenAI | 95.4% | $13.44 | 87 |
| 2 | Claude Opus 4.7 (xhigh) | Anthropic | 89.3% | $23.56 | 256 |
| 3 | Claude Sonnet 4.6 | Anthropic | 82.8% | $32.57 | 443 |
| 4 | Claude Opus 4.7 | Anthropic | 81.6% | $10.35 | 164 |
| 5 | Claude Opus 4.6 | Anthropic | 81.3% | $12.37 | 236 |
| 6 | GPT 5.4 | OpenAI | 54.1% | $0.32 | 12 |
| 7 | Gemini 3 Flash | 51.9% | $0.45 | 76 | |
| 8 | GPT 5.5 | OpenAI | 45.1% | $1.07 | 14 |
| 9 | Gemini 3.1 Pro | 36.0% | $2.46 | 122 | |
| 10 | Claude Haiku 4.5 | Anthropic | 34.1% | $0.87 | 92 |
| 11 | GPT 5 mini | OpenAI | 7.3% | $0.02 | 13 |
| 12 | GPT 5.4 mini | OpenAI | 0.0% | $0.04 | 9 |
| 13 | GPT 5.5 (high) | OpenAI | n/a | $3.22 | 33 |