297generated behavioral tests
71.7%official best score
69.4%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 | 69.4% | ok | 206/297 | $6.81 | 116 | 0.30h | manifest · eval json · eval summary · usage audit |
| 2 | GPT 5.5 (xhigh) | 20260518T232006Z |
Codex /goal | Paper prompt + /goal no internet | 40.7% | ok | 121/297 | n/a | 44 | 0.14h | not exported |
| 3 | GPT 5.5 (xhigh) | 20260519T215046Z |
Codex /goal | Paper prompt + Goal contract no internet | 39.4% | ok | 117/297 | n/a | 68 | 0.19h | 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
69.4% from 206/297 ProgramBench-scored tests. Raw public eval statuses: failure: 91, passed: 215, skipped: 1.
Agent trace summary unavailable.
Likely miss class: behavioral mismatch.
06dabfabaea7tests.test_clone_installer.test_installer_creates_all_expected_dirs (failure)06dabfabaea7tests.test_config.test_invalid_formatter_error (failure)06dabfabaea7tests.test_config.test_config_missing_editor_field (failure)06dabfabaea7tests.test_completions.test_empty_shell_name_error (failure)06dabfabaea7tests.test_config.test_invalid_colorize_type_error (failure)06dabfabaea7tests.test_clone_installer.test_installer_clone_yes (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 | 71.7% | $9.93 | 81 |
| 2 | GPT 5.5 (high) | OpenAI | 66.7% | $4.69 | 60 |
| 3 | Claude Opus 4.6 | Anthropic | 59.9% | $5.80 | 198 |
| 4 | GPT 5.5 | OpenAI | 57.6% | $0.95 | 14 |
| 5 | Claude Sonnet 4.6 | Anthropic | 50.5% | $14.36 | 373 |
| 6 | Claude Opus 4.7 | Anthropic | 49.2% | $3.11 | 86 |
| 7 | GPT 5.4 | OpenAI | 45.1% | $0.40 | 10 |
| 8 | Gemini 3.1 Pro | 22.9% | $1.20 | 70 | |
| 9 | Claude Haiku 4.5 | Anthropic | 22.6% | $0.91 | 159 |
| 10 | Claude Opus 4.7 (xhigh) | Anthropic | 14.1% | $16.68 | 242 |
| 11 | GPT 5.4 mini | OpenAI | 9.4% | $0.04 | 7 |
| 12 | GPT 5 mini | OpenAI | 5.7% | $0.03 | 14 |
| 13 | Gemini 3 Flash | 3.0% | $0.33 | 110 |