CursorBench 3.1 evaluates AI agents on ambiguous, multi-file tasks based on real Cursor sessions, with scores indicating performance. Fable 5 Max achieved the highest score of 72.9%, while GPT-5.5 Extra High scored 64.3%.

Major AI models were tested with the same political, economic, and societal questions while web search was disabled. The results were mapped to reveal the inherent biases of each model, highlighting their leanings in relation to charged topics.

Automated scanning reveals that top AI models frequently achieve high benchmark scores that do not accurately reflect their capabilities. The reliance on these benchmarks has led to a misrepresentation of model performance in the AI industry.

OpenClaw Arena provides a public benchmark to assess AI agents' ability to complete real workflows. Users can compare model performance and cost-effectiveness on actual agent tasks.