224generated behavioral tests
62.1%official best score
58.0%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 | 58.0% | ok | 130/224 | $3.31 | 56 | 0.32h | manifest · eval json · eval summary · usage audit |
| 2 | GPT 5.5 (xhigh) | 20260519T215046Z |
Codex /goal | Paper prompt + Goal contract no internet | 37.9% | ok | 85/224 | n/a | 68 | 0.18h | not exported |
| 3 | GPT 5.5 (xhigh) | 20260518T232006Z |
Codex /goal | Paper prompt + /goal no internet | 25.4% | ok | 57/224 | n/a | 46 | 0.12h | 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
58.0% from 130/224 ProgramBench-scored tests. Raw public eval statuses: failure: 109, passed: 189, skipped: 13.
Agent trace summary unavailable.
Likely miss class: behavioral mismatch, interactive terminal behavior mismatch, terminal rendering mismatch.
e2bdc309a3c0eval.tests.test_basic.test_metadata_lines_ignored (failure)2b31a0cff240tests.test_harvest.test_fixture_top_view[invalid-lines] (failure)2b31a0cff240tests.test_harvest.test_fixture_top_view[recursive] (failure)2b31a0cff240tests.test_harvest.test_sorted_flag (failure)2b31a0cff240tests.test_harvest.test_recursive_stacks (failure)2b31a0cff240tests.test_harvest.test_invalid_lines_handling (failure)
GPT 5.5 (xhigh) · 20260519T215046Z
Public eval evidence has not been exported for this row yet.
GPT 5.5 (xhigh) · 20260518T232006Z
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 (high) | OpenAI | 62.1% | $2.47 | 49 |
| 2 | Claude Opus 4.6 | Anthropic | 59.4% | $15.81 | 252 |
| 3 | GPT 5.5 (xhigh) | OpenAI | 57.6% | $4.30 | 48 |
| 4 | Claude Sonnet 4.6 | Anthropic | 57.6% | $43.26 | 490 |
| 5 | Claude Opus 4.7 | Anthropic | 57.1% | $1.38 | 62 |
| 6 | GPT 5.5 | OpenAI | 55.8% | $0.95 | 14 |
| 7 | Gemini 3 Flash | 44.2% | $0.19 | 57 | |
| 8 | Claude Haiku 4.5 | Anthropic | 41.5% | $0.82 | 157 |
| 9 | GPT 5.4 | OpenAI | 21.9% | $0.07 | 7 |
| 10 | GPT 5.4 mini | OpenAI | 17.0% | $0.01 | 8 |
| 11 | Gemini 3.1 Pro | 14.7% | $2.25 | 77 | |
| 12 | GPT 5 mini | OpenAI | 13.8% | $0.02 | 11 |
| 13 | Claude Opus 4.7 (xhigh) | Anthropic | 1.3% | $5.90 | 135 |