1410generated behavioral tests
84.7%official best score
43.8%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) | 20260518T232006Z |
Codex /goal | Paper prompt + /goal no internet | 43.8% | ok | 618/1410 | n/a | 72 | 0.21h | not exported |
| 2 | GPT 5.5 (xhigh) | 20260519T215046Z |
Codex /goal | Paper prompt + Goal contract no internet | 41.4% | ok | 584/1410 | n/a | 57 | 0.20h | not exported |
| 3 | GPT 5.5 (xhigh) | 20260517T094700Z |
Codex /goal | Mini-SWE-compatible no internet | 40.9% | ok | 576/1410 | $5.48 | 118 | 0.25h | 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
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.
GPT 5.5 (xhigh) · 20260517T094700Z
40.9% from 576/1410 ProgramBench-scored tests. Raw public eval statuses: failure: 112, not_run: 773, passed: 1151.
Agent trace summary unavailable.
Likely miss class: behavioral mismatch.
3dff168fbdedeval.tests.test_args_edge_cases.test_filters_help_output (failure)3dff168fbdedeval.tests.test_advanced_coverage.test_test_mode_with_verbose (failure)3dff168fbdedeval.tests.test_basic.test_filters_help (failure)3dff168fbdedeval.tests.test_advanced_coverage.test_no_sync_option (failure)3dff168fbdedeval.tests.test_args_and_options.test_multiple_verbose_flags (failure)3dff168fbdedeval.tests.test_advanced_coverage.test_filters_string_syntax (failure)
Official ProgramBench Results by Model
| # | Model | Provider | Score | Cost | Calls |
|---|---|---|---|---|---|
| 1 | Claude Opus 4.7 (xhigh) | Anthropic | 84.7% | $27.85 | 344 |
| 2 | GPT 5.5 (high) | OpenAI | 79.7% | $4.26 | 43 |
| 3 | GPT 5.5 (xhigh) | OpenAI | 44.7% | $12.08 | 82 |
| 4 | GPT 5.5 | OpenAI | 41.4% | $1.30 | 20 |
| 5 | Claude Opus 4.7 | Anthropic | 36.0% | $3.27 | 94 |
| 6 | GPT 5.4 | OpenAI | 35.3% | $0.32 | 12 |
| 7 | Claude Haiku 4.5 | Anthropic | 30.1% | $0.45 | 119 |
| 8 | Gemini 3.1 Pro | 28.2% | $1.46 | 98 | |
| 9 | GPT 5.4 mini | OpenAI | 26.5% | $0.02 | 8 |
| 10 | Gemini 3 Flash | 23.8% | $0.33 | 93 | |
| 11 | Claude Sonnet 4.6 | Anthropic | 8.8% | $24.28 | 507 |
| 12 | GPT 5 mini | OpenAI | 5.6% | $0.06 | 18 |
| 13 | Claude Opus 4.6 | Anthropic | 0.0% | $1.36 | 189 |