Pareto frontier · multi-benchmark

AI Pareto Explorer

Pick any two axes across ~75 benchmarks — the frontier recomputes live, with source, configuration, uncertainty, and evaluation age preserved wherever published.

Benchmark X Y Presets

Released Labs

all

Model weights best configuration only frontier only

Data & API

The complete validated snapshot and smaller filtered endpoints are public, CORS-enabled, and documented for scripts, crawlers, and agents.

JSON snapshot API index OpenAPI quality llms.txt

Coverage