Leaderboard

Examine overall author performance across every kernel definition and workload.

Click on the chart to pin a different p and see the ranking at that threshold.

fastp

gemini-2.5-p...gpt-o3gpt-5-2025-0...claude-opus-...
p = 0.95
n = 660 workloads
P range
0.10×1.00×6.06×

Models

Explore model architectures and their kernel implementations

Loading kernels…