710generated behavioral tests
97.3%official best score
95.1%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 | 95.1% | ok | 675/710 | $5.20 | 90 | 0.28h | manifest · eval json · eval summary · usage audit |
| 2 | GPT 5.5 (xhigh) | 20260518T232006Z |
Codex /goal | Paper prompt + /goal no internet | 89.9% | ok | 638/710 | n/a | 51 | 0.14h | not exported |
| 3 | GPT 5.5 (xhigh) | 20260519T215046Z |
Codex /goal | Paper prompt + Goal contract no internet | 76.1% | ok | 540/710 | n/a | 61 | 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
95.1% from 675/710 ProgramBench-scored tests. Raw public eval statuses: failure: 36, passed: 742.
Agent trace summary unavailable.
Likely miss class: behavioral mismatch.
2cb9872af9d3eval.tests.test_edge_cases.test_for_newline_separated (failure)2cb9872af9d3eval.tests.test_error_handling.test_for_with_newlines (failure)2cb9872af9d3eval.tests.test_iteration.test_for_with_spaces (failure)2cb9872af9d3eval.tests.test_iteration.test_stdin_piped (failure)a11a11798c65eval.tests.test_for_items.test_for_with_spaces (failure)a11a11798c65eval.tests.test_regex_intensive.test_regex_with_whitespace_class (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 | 97.3% | $4.91 | 66 |
| 2 | Claude Opus 4.6 | Anthropic | 94.6% | $8.74 | 301 |
| 3 | Claude Opus 4.7 (xhigh) | Anthropic | 94.4% | $10.93 | 196 |
| 4 | GPT 5.5 (high) | OpenAI | 83.0% | $4.57 | 50 |
| 5 | Gemini 3.1 Pro | 81.7% | $1.66 | 176 | |
| 6 | Claude Opus 4.7 | Anthropic | 79.3% | $5.22 | 143 |
| 7 | GPT 5.5 | OpenAI | 55.4% | $1.07 | 18 |
| 8 | Claude Haiku 4.5 | Anthropic | 54.9% | $0.94 | 145 |
| 9 | GPT 5.4 | OpenAI | 52.0% | $0.24 | 12 |
| 10 | Gemini 3 Flash | 52.0% | $0.51 | 142 | |
| 11 | Claude Sonnet 4.6 | Anthropic | 45.8% | $7.56 | 370 |
| 12 | GPT 5 mini | OpenAI | 24.2% | $0.02 | 15 |
| 13 | GPT 5.4 mini | OpenAI | 11.0% | $0.04 | 8 |