293generated behavioral tests
89.4%official best score
94.5%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 | 94.5% | ok | 277/293 | $5.35 | 102 | 0.26h | manifest · eval json · eval summary · usage audit |
| 2 | GPT 5.5 (xhigh) | 20260518T232006Z |
Codex /goal | Paper prompt + /goal no internet | 75.1% | ok | 220/293 | n/a | 46 | 0.14h | not exported |
| 3 | GPT 5.5 (xhigh) | 20260519T215046Z |
Codex /goal | Paper prompt + Goal contract no internet | 68.9% | ok | 202/293 | n/a | 35 | 0.14h | 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
94.5% from 277/293 ProgramBench-scored tests. Raw public eval statuses: failure: 20, passed: 412.
Agent trace summary unavailable.
Likely miss class: behavioral mismatch.
c248a8a914ebtests.test_output_formats.test_verbose_and_junit_conflict (failure)34ab4782d13eeval.tests.test_connection_options.test_dns_resolver_flag (failure)34ab4782d13eeval.tests.test_connection_options.test_proxy_flag (failure)d6003e5f008btests.test_http_features.test_multiple_non_contiguous_ranges (failure)d6003e5f008btests.test_filtering.test_scheme_filtering_non_http_schemes_ignored (failure)d6003e5f008btests.test_link_checking.test_connection_refused_error_handling (failure)
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.
Official ProgramBench Results by Model
| # | Model | Provider | Score | Cost | Calls |
|---|---|---|---|---|---|
| 1 | GPT 5.5 (xhigh) | OpenAI | 89.4% | $10.53 | 90 |
| 2 | Claude Sonnet 4.6 | Anthropic | 88.1% | $11.78 | 290 |
| 3 | Claude Opus 4.7 | Anthropic | 87.4% | $3.77 | 111 |
| 4 | Claude Opus 4.7 (xhigh) | Anthropic | 85.7% | $2.53 | 70 |
| 5 | GPT 5.5 (high) | OpenAI | 84.0% | $2.24 | 26 |
| 6 | GPT 5.4 | OpenAI | 81.2% | $0.30 | 13 |
| 7 | Claude Opus 4.6 | Anthropic | 80.5% | $7.24 | 170 |
| 8 | GPT 5.5 | OpenAI | 78.8% | $1.37 | 19 |
| 9 | Gemini 3.1 Pro | 60.8% | $1.03 | 60 | |
| 10 | Gemini 3 Flash | 59.7% | $0.33 | 79 | |
| 11 | Claude Haiku 4.5 | Anthropic | 46.1% | $0.50 | 87 |
| 12 | GPT 5.4 mini | OpenAI | 33.4% | $0.03 | 9 |
| 13 | GPT 5 mini | OpenAI | 27.3% | $0.02 | 14 |