List your company

Bongard Problem Verification on Bongard Problems BPs #2–#100 (test)

86.5AccuracyClaude 3.7 Sonnet
72.14875.87479.683.326Feb 3, 2026

Evaluation results

MethodLinks
Claude 3.7 Sonnetverifier=natural language + programs2026.0286.5
Claude 3.7 Sonnetverifier=natural language2026.0283.5
Claude 3.7 Sonnetverifier=programs2026.0282.1
GPT-4overifier=natural language + programs2026.0279
GPT-4overifier=natural language2026.0277.5
GPT-4overifier=programs2026.0272.7