agents benchmark
KernelBench Hard Leaderboard
KernelBench Hard evaluates agentic GPU kernel optimization on the hardest problem set. Each question is scored by the agent's submitted operator TFLOPs relative to the theoretical peak of the current hardware, with the benchmark score being the average across all questions.
Updated Aug 17, 2026
Models1
Model coverage1
MetricScore
EvidenceB
KernelBench Hard Ranking
Higher score ranks better on this benchmark.
Sort by
Rank | Model | Score | Percentile | Participants | Evidence | Evaluated |
|---|
| Rank01 | ModelMI | Score28.8% | Percentile100.0% | Participants1 | EvidenceC | Evaluated |
KernelBench Hard Highlights
The leading models and scores on this benchmark.
The Top AI Models for KernelBench Hard
The first five results on this benchmark, with official price and output speed added where the model identity can be matched.
Ranking basisThis kernelbench hard AI model leaderboard uses descending score in the benchmark's original unit. The leaderboard ranking keeps matched price and speed data separate from benchmark evidence.
- 01MIMiniMax M3MiniMax
- Score
- 28.8%
- Price
- $0.30 input / $1.2 output per 1M tokens
- Speed
- Up to 214 tok/s via Together
Strengths
- Ranks #1 of 1 compared models
- 100th percentile on this benchmark
- C evidence result
Considerations
- This result measures KernelBench Hard, not total model capability
Selection summary
Best AI Models for KernelBench Hard
MiniMax M3 currently leads KernelBench Hard with 28.8%. 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.
What is KernelBench Hard?
What KernelBench Hard measures and how its scores work.
KernelBench Hard evaluates agentic GPU kernel optimization on the hardest problem set. Each question is scored by the agent's submitted operator TFLOPs relative to the theoretical peak of the current hardware, with the benchmark score being the average across all questions.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.
- Family
- KernelBench Hard
- Modality
- text
- Primary category
- agents
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- kernelbench-hard|llm-stats-current
Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about KernelBench Hard.
Which model scores highest on KernelBench Hard?
MiniMax M3 is currently ranked first with 28.8%.
What does KernelBench Hard measure?
KernelBench Hard evaluates agentic GPU kernel optimization on the hardest problem set. Each question is scored by the agent's submitted operator TFLOPs relative to the theoretical peak of the current hardware, with the benchmark score being the average across all questions.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
1 model results are currently shown.
Does this benchmark affect the overall score?
No. This benchmark is shown for reference but does not contribute to the overall score.