safety benchmark
PatchBench evaluates whether a model can produce patches that fix security vulnerabilities while preserving intended program behavior.
Updated Nov 22, 2026
Higher score ranks better on this benchmark.
Rank | Model | Score | Percentile | Participants | Evidence | Evaluated |
|---|
| Rank01 | ModelOP | Score79.70% | Percentile100.00% | Participants2 | EvidenceC | Evaluated |
| Rank02 | ModelOP | Score75.60% | Percentile0.00% | Participants2 | EvidenceC | Evaluated |
The leading models and scores on this benchmark.
A closer view of the leading scores on this benchmark.
The first five results on this benchmark, with official price and output speed added where the model identity can be matched.
What PatchBench measures and how its scores work.
PatchBench is a safety benchmark reported in OpenAI's GPT-6.2 launch.
It measures Score, expressed as a ratio, for producing patches that fix security vulnerabilities while preserving intended program behavior.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.
LLM Stats. Benchmark scores retain their original unit. Overall score eligibility is shown separately.
Common questions about PatchBench.
GPT-6.2 Astra is currently ranked first with 79.70%.
The current leaders are GPT-6.2 Astra (79.70%) and GPT-6.2 Sol (75.60%).
GPT-6.2 Sol has the lowest matched official input price at $2.0 input / $10 output per 1M tokens.
No matched runtime record is currently available.
No. This benchmark measures one defined capability or task. The overall LLMBoard score uses a separate aggregation across eligible benchmark evidence.
It measures Score, expressed as a ratio, for producing patches that fix security vulnerabilities while preserving intended program behavior.
Yes. Higher values rank better for this benchmark.
2 model results are currently shown.
No. This benchmark is shown for reference but does not contribute to the overall score.
Ranking basisThis patchbench AI model leaderboard uses descending score in the benchmark original unit. The leaderboard ranking keeps matched price and speed data separate from benchmark evidence.
Selection summary
GPT-6.2 Astra currently leads PatchBench with 79.70%. It is the top model on this specific benchmark, while the best LLM for the broader task should also be checked against other benchmarks, price, and runtime.
Use this leaderboard with the supporting benchmark results and coverage details above. A leaderboard position summarizes the selected ranking signal; it does not replace workload-specific testing.