KernelBench Hub - Mega — leaderboard
Metric: Best Speedup vs Reference (x). Source: kernelbench.com. 12 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Fable 5 | 24.61 |
| 2 | Claude Opus 4.8 | 19.35 |
| 3 | Kimi K3 | 18.09 |
| 4 | GLM-5.2 | 11.14 |
| 5 | MiniMax-M3 | 4.15 |
| 6 | Claude Sonnet 5 | 4.03 |
| 7 | GPT-5.6 Sol | 2.64 |
| 8 | Kimi K2.7 Code | 2.59 |
| 9 | DeepSeek V4 Pro | 2.11 |
| 10 | Grok 4.5 | 0.82 |
Interactive version: theaggregate.ai/benchmark?slug=kernelbench-hub-mega · How the rankings work · Data refreshed daily, snapshot 2026-07-22.