2000generated behavioral tests
39.5%official best score
34.5%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 | 34.5% | ok | 690/2000 | $4.29 | 82 | 0.22h | manifest · eval json · eval summary · usage audit |
| 2 | GPT 5.5 (xhigh) | 20260519T215046Z |
Codex /goal | Paper prompt + Goal contract no internet | 25.6% | ok | 511/2000 | n/a | 44 | 0.17h | not exported |
| 3 | GPT 5.5 (xhigh) | 20260518T232006Z |
Codex /goal | Paper prompt + /goal no internet | 1.1% | ok | 23/2000 | n/a | 43 | 0.15h | 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
34.5% from 690/2000 ProgramBench-scored tests. Raw public eval statuses: failure: 1342, passed: 704.
Agent trace summary unavailable.
Likely miss class: behavioral mismatch, exact version/output mismatch.
556723988b33eval.tests.test_advanced_features.test_sh_encode_operator (failure)556723988b33eval.tests.test_advanced_features.test_eval_all_multiple_files (failure)556723988b33eval.tests.test_basic_invocation.test_help_eval_all_subcommand (failure)556723988b33eval.tests.test_basic_invocation.test_exit_status_flag_no_match (failure)556723988b33eval.tests.test_advanced_features.test_strload_file (failure)556723988b33eval.tests.test_advanced_features.test_eval_expression_string (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 | Claude Sonnet 4.6 | Anthropic | 39.5% | $35.51 | 532 |
| 2 | GPT 5.5 (xhigh) | OpenAI | 37.5% | $9.50 | 87 |
| 3 | GPT 5.5 (high) | OpenAI | 29.6% | $2.60 | 32 |
| 4 | Claude Opus 4.7 (xhigh) | Anthropic | 16.5% | $5.49 | 92 |
| 5 | Claude Haiku 4.5 | Anthropic | 16.0% | $1.55 | 183 |
| 6 | GPT 5.5 | OpenAI | 14.5% | $1.00 | 13 |
| 7 | Claude Opus 4.7 | Anthropic | 11.5% | $1.28 | 38 |
| 8 | Gemini 3.1 Pro | 10.3% | $1.02 | 60 | |
| 9 | GPT 5.4 | OpenAI | 6.9% | $0.76 | 20 |
| 10 | GPT 5.4 mini | OpenAI | 5.1% | $0.02 | 6 |
| 11 | GPT 5 mini | OpenAI | 4.0% | $0.02 | 8 |
| 12 | Gemini 3 Flash | 1.1% | $0.19 | 65 | |
| 13 | Claude Opus 4.6 | Anthropic | 0.0% | $2.85 | 217 |