Every model's score on every benchmark we track, each linked to where it was published. Independent runs and lab-reported numbers are kept apart. Many of these benchmarks feed the Respan Index.
177 of 177 benchmarks
Agents
13 models, higher is better
Tests computer-use agents on real tasks across desktop operating systems and apps, scored by whether each task ends in the right state.
Benchmark data from llmmetric.com, Respan's model data service.