1114generated behavioral tests
72.5%official best score
67.1%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) | 20260518T232006Z |
Codex /goal | Paper prompt + /goal no internet | 67.1% | ok | 748/1114 | n/a | 60 | 0.19h | not exported |
| 2 | GPT 5.5 (xhigh) | 20260519T215046Z |
Codex /goal | Paper prompt + Goal contract no internet | 64.0% | ok | 713/1114 | n/a | 79 | 0.23h | not exported |
| 3 | GPT 5.5 (xhigh) | 20260517T094700Z |
Codex /goal | Mini-SWE-compatible no internet | 4.1% | ok | 46/1114 | $3.72 | 43 | 0.16h | manifest · eval json · eval summary · usage audit |
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) · 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.
GPT 5.5 (xhigh) · 20260517T094700Z
4.1% from 46/1114 ProgramBench-scored tests. Raw public eval statuses: error: 6, failure: 1211, not_run: 23, passed: 80, skipped: 6.
Agent trace summary unavailable.
Likely miss class: behavioral mismatch, exact version/output mismatch.
d044b8e852e1tests.test_advanced_features.test_init_in_existing_directory_with_files (failure)d044b8e852e1tests.test_basic_invocation.test_capital_v_flag (failure)d044b8e852e1tests.test_book_loading.test_book_without_book_toml (failure)d044b8e852e1tests.test_advanced_features.test_completions_output_validity (failure)d044b8e852e1tests.test_advanced_features.test_build_with_configuration_options (failure)d044b8e852e1tests.test_basic_invocation.test_help_for_serve (failure)
Official ProgramBench Results by Model
| # | Model | Provider | Score | Cost | Calls |
|---|---|---|---|---|---|
| 1 | GPT 5.5 (xhigh) | OpenAI | 72.5% | $11.95 | 87 |
| 2 | GPT 5.5 (high) | OpenAI | 61.8% | $3.88 | 33 |
| 3 | Claude Sonnet 4.6 | Anthropic | 55.5% | $50.89 | 644 |
| 4 | GPT 5.5 | OpenAI | 48.8% | $1.10 | 15 |
| 5 | GPT 5.4 | OpenAI | 40.8% | $0.26 | 9 |
| 6 | Claude Opus 4.7 | Anthropic | 37.9% | $1.52 | 60 |
| 7 | Claude Opus 4.7 (xhigh) | Anthropic | 31.2% | $5.72 | 155 |
| 8 | Gemini 3 Flash | 25.6% | $0.25 | 80 | |
| 9 | Gemini 3.1 Pro | 21.5% | $1.83 | 110 | |
| 10 | Claude Haiku 4.5 | Anthropic | 18.2% | $0.74 | 139 |
| 11 | GPT 5 mini | OpenAI | 8.6% | $0.04 | 18 |
| 12 | Claude Opus 4.6 | Anthropic | 4.1% | $13.14 | 244 |
| 13 | GPT 5.4 mini | OpenAI | 0.0% | $0.03 | 9 |