Benchmarks
347 sources · 1,258 models × 3,038 benchmarks · 158,912 scoresKernelBench Mega evaluates agent-generated kernels on a two-problem GPU optimization deck. This suite-level row defines the family and has no direct score observations; hardware-specific child benchmarks report the best valid Kimi-Linear decode speedup.
No signal or difficulty estimate. Held out of the fit by design.
This is a benchmark-family page. Scores are recorded on the component benchmarks below rather than directly on the parent, because the components use different score scales and directions.
| Metric | Models | Leader | Direction | Date |
|---|---|---|---|---|
| KernelBench Mega — RTX PRO 6000 Best Decode Speedup | 12 | Claude Fable 518.7 | higher is better | 2026-07-17 |
| KernelBench Mega — B200 Best Decode Speedup | 6 | Opus 4.819.4 | higher is better | 2026-07-17 |
| KernelBench Mega — H100 Best Decode Speedup | 5 | Opus 4.815.5 | higher is better | 2026-07-17 |