Skip to main content
Evaluation sources/BullshitBench v2
ParticipatingKnowledge & language

BullshitBench v2

BullshitBench · Public model evaluation results from BullshitBench v2.

Official evaluation
Source description

This source measures concrete delivery performance in Knowledge & language; raw scores are shown separately from the normalized capability score.

Limitation: Different versions, reasoning settings, and test sets may change scores; read them with the data date.

How it is used

Participating sources affect capability scores; reference, cross-reference, and observing sources are for comparison and verification only.

Model results from this source

Qwen3.8 Max0.9580.12026-11-05
Claude Opus 4.80.9580.12026-11-05
Claude Sonnet 4.60.9177.82026-11-05
Claude Opus 4.50.977.22026-11-05
Claude Opus 4.60.8775.42026-11-05
Claude Opus 4.70.8373.12026-11-05
Claude Sonnet 50.871.42026-11-05
Claude Sonnet 4.50.7970.82026-11-05
Qwen3.5 397B-A17B0.7870.22026-11-05
Claude Fable 5.10.7769.72026-11-05
Claude Haiku 4.50.7769.72026-11-05
Kimi K30.7467.92026-11-05
Claude Opus 50.7367.32026-11-05
Qwen3.6 Plus0.7266.82026-11-05
Qwen3.7 Max0.7266.82026-11-05
GLM 5.30.7266.82026-11-05
GPT-6 Astra0.6965.02026-11-05
Gemini 3.5 Flash Lite0.6663.32026-11-05
Gemini 4 Argon0.6663.32026-10-18
Gemini 4 Argon0.6663.32026-11-05
Gemini 4 Argon0.6663.32026-10-17
Grok 4.60.6663.32026-11-05
Kimi K2.60.6562.72026-11-05
GPT-6.1 Sol0.6562.72026-11-05

Page 1 of 4

Frequently asked questions