718generated behavioral tests
81.5%official best score
75.2%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 | 75.2% | ok | 540/718 | $3.74 | 77 | 0.22h | manifest · eval json · eval summary · usage audit |
| 2 | GPT 5.5 (xhigh) | 20260519T215046Z |
Codex /goal | Paper prompt + Goal contract no internet | 53.5% | ok | 384/718 | n/a | 78 | 0.27h | not exported |
| 3 | GPT 5.5 (xhigh) | 20260518T232006Z |
Codex /goal | Paper prompt + /goal no internet | 51.7% | ok | 371/718 | n/a | 76 | 0.20h | 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
75.2% from 540/718 ProgramBench-scored tests. Raw public eval statuses: failure: 193, passed: 915.
Agent trace summary unavailable.
Likely miss class: behavioral mismatch.
f990797ddcbaeval.tests.test_extensions_prefixes.test_extension_file_not_found (failure)f990797ddcbaeval.tests.test_filtering_validation.test_invalid_code_blacklist (failure)f990797ddcbaeval.tests.test_extensions_prefixes.test_prefix_file_not_found (failure)f990797ddcbaeval.tests.test_extensions_prefixes.test_empty_extension_file (failure)f990797ddcbaeval.tests.test_proxy_config.test_invalid_proxy_url (failure)f990797ddcbaeval.tests.test_filtering_validation.test_invalid_length_format (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 (xhigh) | OpenAI | 81.5% | $7.31 | 54 |
| 2 | GPT 5.5 (high) | OpenAI | 73.8% | $3.20 | 38 |
| 3 | Claude Opus 4.7 (xhigh) | Anthropic | 70.8% | $6.03 | 98 |
| 4 | Claude Opus 4.6 | Anthropic | 66.7% | $5.89 | 95 |
| 5 | GPT 5.5 | OpenAI | 63.9% | $0.91 | 14 |
| 6 | Claude Sonnet 4.6 | Anthropic | 51.8% | $15.98 | 345 |
| 7 | GPT 5.4 | OpenAI | 51.7% | $0.39 | 11 |
| 8 | Gemini 3 Flash | 51.5% | $0.31 | 80 | |
| 9 | Gemini 3.1 Pro | 46.8% | $0.85 | 33 | |
| 10 | Claude Opus 4.7 | Anthropic | 29.7% | $2.28 | 71 |
| 11 | GPT 5 mini | OpenAI | 27.2% | $0.02 | 9 |
| 12 | Claude Haiku 4.5 | Anthropic | 25.5% | $0.64 | 111 |
| 13 | GPT 5.4 mini | OpenAI | 24.7% | $0.04 | 9 |