ModelRadar
/

Benchmark research

Describe what the model has to do. Hard constraints (memory, license, modality) are applied first, then ranking by benchmark results weighted by how much each benchmark is still worth.