2549generated behavioral tests
65.8%official best score
60.0%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 | 60.0% | ok | 1530/2549 | $4.86 | 80 | 0.23h | manifest · eval json · eval summary · usage audit |
| 2 | GPT 5.5 (xhigh) | 20260519T215046Z |
Codex /goal | Paper prompt + Goal contract no internet | 59.8% | ok | 1524/2549 | n/a | 53 | 0.18h | not exported |
| 3 | GPT 5.5 (xhigh) | 20260518T232006Z |
Codex /goal | Paper prompt + /goal no internet | 53.6% | ok | 1367/2549 | n/a | 56 | 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
60.0% from 1530/2549 ProgramBench-scored tests. Raw public eval statuses: failure: 1289, passed: 2068.
Agent trace summary unavailable.
Likely miss class: behavioral mismatch, exact version/output mismatch.
e874f7a39dcdtests.test_absolute_final.test_ultimate_binary_all_outputs (failure)e874f7a39dcdtests.test_actual_scans.test_actual_scan_with_config_file (failure)e874f7a39dcdtests.test_coverage_boost_final.test_echo_pcap_options (failure)e874f7a39dcdtests.test_coverage_final_push.test_sL_mode_cidr (failure)e874f7a39dcdtests.test_advanced.test_ping_option (failure)e874f7a39dcdtests.test_basic_fixed.test_echo_with_exclude_ip (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 | GPT 5.5 (xhigh) | OpenAI | 65.8% | $8.94 | 74 |
| 2 | Claude Opus 4.6 | Anthropic | 57.0% | $13.59 | 295 |
| 3 | GPT 5.5 (high) | OpenAI | 53.7% | $5.02 | 40 |
| 4 | Gemini 3.1 Pro | 52.6% | $1.58 | 100 | |
| 5 | Gemini 3 Flash | 52.3% | $0.26 | 85 | |
| 6 | Claude Opus 4.7 | Anthropic | 50.6% | $1.13 | 49 |
| 7 | Claude Sonnet 4.6 | Anthropic | 48.8% | $14.79 | 500 |
| 8 | GPT 5.5 | OpenAI | 46.3% | $1.30 | 19 |
| 9 | GPT 5.4 | OpenAI | 40.7% | $0.24 | 9 |
| 10 | Claude Opus 4.7 (xhigh) | Anthropic | 38.3% | $4.48 | 109 |
| 11 | Claude Haiku 4.5 | Anthropic | 30.3% | $1.65 | 197 |
| 12 | GPT 5 mini | OpenAI | 21.3% | $0.02 | 15 |
| 13 | GPT 5.4 mini | OpenAI | 18.5% | $0.03 | 8 |