List your company

Video Generation Reasoning on VideoThinkBench (test)

44.7Eyeballing-Point AccuracySora-2
27.12431.68736.2540.813Nov 6, 2025

Evaluation results

MethodLinks
Sora-2Eyeballing-puzzles-evaluation=Major Frame, Text-centric-tasks-evaluation=audio2025.1144.73834.581.951.957.51.313.340.468.665.361.279.168.649.8
Claude Sonnet 4.52025.1136.226.350.580.169.9825.3043.89086.365.682.381.156.2
GPT-5 high2025.1133.62432.598.566.7850.5042.697.285.269.680.683.256.1
Gemini 2.5 Pro2025.1127.82134.594.983.7671.9041.394.884.566.78382.355