agents benchmark
KernelGen 1P
KernelGen 1P is an OpenAI AI-self-improvement evaluation that measures whether models can write and optimize compute kernels, part of the suite tracking progress toward accelerating internal research.
Updated Aug 7, 2026
Published models3
Registry coverage3
MetricScore
EvidenceB
KernelGen 1P leaderboard
Sorted by the source-provided rank. Higher score is better according to the registry.
| 1 | OP | 61.1% | 100.0% | 3 | C | |
| 2 | OP | 49.2% | 50.0% | 3 | C | |
| 3 | OP | 22.4% | 0.0% | 3 | C |
Score distribution
Top published rows on the benchmark's original scale.
KernelGen 1P highlights
The top published results on this benchmark's own scale.
What is KernelGen 1P?
Definition and scoring fields from the benchmark registry.
KernelGen 1P is an OpenAI AI-self-improvement evaluation that measures whether models can write and optimize compute kernels, part of the suite tracking progress toward accelerating internal research.
Scores are shown in ratio. The current registry marks this benchmark as not independently verified with evidence level B.
- Family
- KernelGen 1P
- Modality
- text
- Primary category
- agents
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- kernelgen-1p|llm-stats-current
Source-native results are preserved. Eligibility for the overall LLMBoard score is a separate policy decision.
FAQ
Common questions about KernelGen 1P.
Which model scores highest on KernelGen 1P?
GPT-5.6 Sol is currently ranked first with 61.1%.
What does KernelGen 1P measure?
KernelGen 1P is an OpenAI AI-self-improvement evaluation that measures whether models can write and optimize compute kernels, part of the suite tracking progress toward accelerating internal research.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
3 unique published model results are currently shown.
Does this benchmark affect the overall score?
This benchmark is preserved as source-native evidence but is not eligible for the current overall score.