383generated behavioral tests
98.7%official best score
95.8%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 | 95.8% | ok | 367/383 | $2.58 | 49 | 0.20h | manifest · eval json · eval summary · usage audit |
| 2 | GPT 5.5 (xhigh) | 20260519T215046Z |
Codex /goal | Paper prompt + Goal contract no internet | 91.9% | ok | 352/383 | n/a | 59 | 0.20h | not exported |
| 3 | GPT 5.5 (xhigh) | 20260518T232006Z |
Codex /goal | Paper prompt + /goal no internet | 91.4% | ok | 350/383 | n/a | 43 | 0.17h | 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
95.8% from 367/383 ProgramBench-scored tests. Raw public eval statuses: failure: 17, passed: 399, skipped: 3.
Agent trace summary unavailable.
Likely miss class: behavioral mismatch, exact version/output mismatch.
0fd941cd858feval.tests.test_pingu.test_empty_host_produces_error (failure)d7a5dbbf1b14tests.test_errors.test_empty_string_hostname (failure)d7a5dbbf1b14tests.test_edge_cases.test_ipv4_all_zeros_resolves_to_localhost (failure)d7a5dbbf1b14tests.test_art_rendering.test_ansi_reset_codes_present (failure)d7a5dbbf1b14tests.test_art_rendering.test_colorize_black_character_replacement (failure)d7a5dbbf1b14tests.test_art_rendering.test_colorize_white_character_replacement (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 Opus 4.7 (xhigh) | Anthropic | 98.7% | $5.81 | 109 |
| 2 | Claude Opus 4.7 | Anthropic | 96.6% | $3.36 | 104 |
| 3 | GPT 5.5 (xhigh) | OpenAI | 96.3% | $4.39 | 45 |
| 4 | GPT 5.5 (high) | OpenAI | 96.1% | $2.73 | 35 |
| 5 | Claude Opus 4.6 | Anthropic | 94.0% | $6.93 | 165 |
| 6 | Claude Sonnet 4.6 | Anthropic | 92.2% | $3.78 | 175 |
| 7 | GPT 5.5 | OpenAI | 88.0% | $0.93 | 19 |
| 8 | Claude Haiku 4.5 | Anthropic | 74.2% | $0.48 | 87 |
| 9 | GPT 5 mini | OpenAI | 66.3% | $0.02 | 18 |
| 10 | GPT 5.4 | OpenAI | 62.1% | $0.22 | 10 |
| 11 | Gemini 3 Flash | 59.3% | $0.30 | 59 | |
| 12 | GPT 5.4 mini | OpenAI | 18.3% | $0.01 | 9 |
| 13 | Gemini 3.1 Pro | 0.0% | $1.11 | 48 |