464generated behavioral tests
47.8%official best score
43.3%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 | 43.3% | ok | 201/464 | n/a | 65 | 0.22h | not exported |
| 2 | GPT 5.5 (xhigh) | 20260517T094700Z |
Codex /goal | Mini-SWE-compatible no internet | 37.9% | ok | 176/464 | $5.82 | 87 | 0.22h | manifest · eval json · eval summary · usage audit |
| 3 | GPT 5.5 (xhigh) | 20260519T215046Z |
Codex /goal | Paper prompt + Goal contract no internet | 35.3% | ok | 164/464 | n/a | 96 | 0.32h | 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) · 20260518T232006Z
Public eval evidence has not been exported for this row yet.
GPT 5.5 (xhigh) · 20260517T094700Z
37.9% from 176/464 ProgramBench-scored tests. Raw public eval statuses: failure: 297, passed: 178, skipped: 1.
Agent trace summary unavailable.
Likely miss class: behavioral mismatch.
06dabfabaea7tests.test_analysis.test_percent_with_character_count_combined (failure)06dabfabaea7tests.test_classification.test_default_xml_marker_detection (failure)06dabfabaea7tests.test_core_counting.test_multilanguage_aggregation_accuracy (failure)06dabfabaea7tests.test_core_counting.test_deeply_nested_directory_5_levels (failure)06dabfabaea7tests.test_classification.test_custom_marker_not_found (failure)06dabfabaea7tests.test_classification.test_no_gen_excludes_only_generated (failure)
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 (high) | OpenAI | 47.8% | $5.71 | 67 |
| 2 | Claude Sonnet 4.6 | Anthropic | 37.7% | $46.35 | 593 |
| 3 | Claude Opus 4.6 | Anthropic | 37.3% | $15.01 | 315 |
| 4 | Claude Opus 4.7 (xhigh) | Anthropic | 36.4% | $20.65 | 267 |
| 5 | GPT 5.5 (xhigh) | OpenAI | 28.2% | $17.09 | 107 |
| 6 | GPT 5.5 | OpenAI | 20.7% | $0.83 | 11 |
| 7 | Claude Opus 4.7 | Anthropic | 19.6% | $1.69 | 54 |
| 8 | Gemini 3.1 Pro | 11.4% | $1.22 | 98 | |
| 9 | Claude Haiku 4.5 | Anthropic | 8.2% | $0.60 | 86 |
| 10 | Gemini 3 Flash | 5.8% | $0.23 | 76 | |
| 11 | GPT 5.4 | OpenAI | 4.5% | $0.26 | 9 |
| 12 | GPT 5 mini | OpenAI | 1.9% | $0.03 | 13 |
| 13 | GPT 5.4 mini | OpenAI | 1.7% | $0.03 | 9 |