Benchmarks
Benchmarks

OSWorld-Verified

Agentsunit % · normalized over [0, 90]

Verified/cleaned subset of OSWorld — real desktop-GUI computer-use tasks scored pass@1.

#ModelSourceScoreNormalized
Score distribution
24 tracked results across the normalization window
090
Score vs. parameters
Open-weights models, log-x params
1B10B100B1000B