KernelBench Hard
We show this table for reference; we do not rank on it.
A display-only benchmark for difficult GPU kernel implementation and optimization tasks.
Benchmark score on KernelBench Hard — September 27, 2026
We compile the KernelBench Hard rows from provider self-reports. MiniMax M3 leads the table at 28.8%. We do not use these results to rank models overall.
1 modelCodingCurrentDisplay onlyUpdated September 27, 2026
Benchmark score table (1 model)
ScoreAbout KernelBench Hard
Year
2026
Tasks
Hard GPU kernel coding tasks
Format
Task success rate
Difficulty
Specialized systems programming
MiniMax reports KernelBench Hard in the M3 comparison chart. BenchLM keeps it separate from Terminal-Bench Hard and excludes it from weighted coding scores.
Freshness and provenance
Version
KernelBench Hard 2026
Refresh cadence
Quarterly
Staleness state
Current
Question availability
Public benchmark set
BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.
Questions
What does KernelBench Hard measure?
A display-only benchmark for difficult GPU kernel implementation and optimization tasks.
Which model scores highest on KernelBench Hard?
MiniMax M3 by MiniMax currently leads with a score of 28.8% on KernelBench Hard.
How many models are evaluated on KernelBench Hard?
1 AI models have been evaluated on KernelBench Hard on BenchLM.
Know when it’s worth switching models
The model to choose, the cheaper alternative, and the release we would wait on.
Read a sample issueJoin 2,000+ readers.
One email each week. Unsubscribe anytime.