reasoning
BullshitBench v2
- #194medium of 1963.0%
Leader: Opus 4.8 · 95.0%
Measured on 4 benchmarks.
General frontier
26.0
#89 of 136 models · best at medium
includes estimates: 5% of the mix is measured for this model
reasoning
Leader: Opus 4.8 · 95.0%
composite
Leader: Gemini 4 · 1525
composite
Leader: Gemini 4 · 1553
other
Leader: Fable 5 · 1308