606generated behavioral tests
90.6%official best score
86.6%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 | 86.6% | ok | 525/606 | $5.83 | 84 | 0.31h | manifest · eval json · eval summary · usage audit |
| 2 | GPT 5.5 (xhigh) | 20260518T232006Z |
Codex /goal | Paper prompt + /goal no internet | 73.9% | ok | 448/606 | n/a | 44 | 0.16h | not exported |
| 3 | GPT 5.5 (xhigh) | 20260519T215046Z |
Codex /goal | Paper prompt + Goal contract no internet | 28.1% | ok | 170/606 | n/a | 52 | 0.13h | 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
86.6% from 525/606 ProgramBench-scored tests. Raw public eval statuses: failure: 101, passed: 681.
Agent trace summary unavailable.
Likely miss class: behavioral mismatch.
324281274037eval.tests.test_diffr.test_large_diff_threshold_zero (failure)324281274037eval.tests.test_diffr.test_integration_combined_options (failure)324281274037eval.tests.test_diffr.test_error_missing_threshold_argument (failure)324281274037eval.tests.test_diffr.test_debug_flag (failure)324281274037eval.tests.test_diffr.test_error_missing_color_value (failure)324281274037eval.tests.test_diffr.test_large_threshold_triggers_skip (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 | 90.6% | $9.34 | 74 |
| 2 | GPT 5.5 (high) | OpenAI | 88.4% | $5.85 | 51 |
| 3 | Claude Opus 4.7 (xhigh) | Anthropic | 87.8% | $27.18 | 193 |
| 4 | Claude Opus 4.7 | Anthropic | 84.7% | $4.69 | 102 |
| 5 | Claude Sonnet 4.6 | Anthropic | 80.7% | $27.33 | 384 |
| 6 | Claude Opus 4.6 | Anthropic | 77.2% | $13.03 | 208 |
| 7 | GPT 5.5 | OpenAI | 74.8% | $0.89 | 15 |
| 8 | Gemini 3.1 Pro | 66.2% | $3.33 | 154 | |
| 9 | GPT 5.4 | OpenAI | 64.5% | $0.19 | 8 |
| 10 | GPT 5 mini | OpenAI | 55.1% | $0.03 | 15 |
| 11 | Claude Haiku 4.5 | Anthropic | 54.6% | $1.11 | 143 |
| 12 | Gemini 3 Flash | 52.3% | $0.25 | 76 | |
| 13 | GPT 5.4 mini | OpenAI | 0.0% | $0.02 | 7 |