The community shapes coverage. Models and benchmarks are suggested by researchers; benchmark.ai selects, sources, and evaluates them independently.
benchmark.ai independently selects, sources, configures, and evaluates every model under fixed, documented settings. You do not need to provide access or generate outputs. Tell us what to add and why it is worth including.
Propose a new modality or capability to evaluate. Accepted proposals are credited in the changelog.
For collaborations, custom benchmark requests, data questions, or press.
