Google
Measured on 1 benchmarks.
General frontier
23.6
#100 of 136 models
includes estimates: 15% of the mix is measured for this model
reasoning
Leader: Opus 5.5 · 88.4%