920generated behavioral tests
82.0%official best score
74.9%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 | 74.9% | ok | 689/920 | n/a | 53 | 0.22h | not exported |
| 2 | GPT 5.5 (xhigh) | 20260517T094700Z |
Codex /goal | Mini-SWE-compatible no internet | 74.2% | ok | 683/920 | $4.59 | 58 | 0.19h | manifest · eval json · eval summary · usage audit |
| 3 | GPT 5.5 (xhigh) | 20260519T215046Z |
Codex /goal | Paper prompt + Goal contract no internet | 64.3% | ok | 592/920 | n/a | 53 | 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) · 20260518T232006Z
Public eval evidence has not been exported for this row yet.
GPT 5.5 (xhigh) · 20260517T094700Z
74.2% from 683/920 ProgramBench-scored tests. Raw public eval statuses: failure: 242, passed: 743.
Agent trace summary unavailable.
Likely miss class: behavioral mismatch.
4b36d953e293eval.tests.test_edge_cases.TestEnvironmentVariables.test_rust_log_override (failure)4b36d953e293eval.tests.test_edge_cases.TestEdgeCases.test_invalid_base_address (failure)4b36d953e293eval.tests.test_advanced.TestCombinedFeatures.test_feature_group_with_peripheral (failure)4b36d953e293eval.tests.test_advanced.TestEnumerations.test_enum_with_defmt (failure)4b36d953e293eval.tests.test_features.TestCodeGenerationFeatures.test_feature_group_flag (failure)4b36d953e293eval.tests.test_features.TestCodeGenerationFeatures.test_feature_peripheral_flag (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 (xhigh) | OpenAI | 82.0% | $21.57 | 127 |
| 2 | Claude Sonnet 4.6 | Anthropic | 72.9% | $25.43 | 389 |
| 3 | GPT 5.5 (high) | OpenAI | 69.0% | $4.94 | 39 |
| 4 | Claude Opus 4.6 | Anthropic | 65.9% | $12.91 | 218 |
| 5 | GPT 5.5 | OpenAI | 61.7% | $1.83 | 28 |
| 6 | Claude Opus 4.7 (xhigh) | Anthropic | 58.0% | $1.81 | 50 |
| 7 | Gemini 3.1 Pro | 45.5% | $1.86 | 117 | |
| 8 | Claude Haiku 4.5 | Anthropic | 30.9% | $0.66 | 92 |
| 9 | Claude Opus 4.7 | Anthropic | 14.7% | $0.61 | 27 |
| 10 | GPT 5.4 | OpenAI | 9.5% | $0.09 | 6 |
| 11 | GPT 5.4 mini | OpenAI | 8.0% | $0.03 | 9 |
| 12 | GPT 5 mini | OpenAI | 7.3% | $0.04 | 19 |
| 13 | Gemini 3 Flash | 0.9% | $0.39 | 82 |