WatchAct - State Change: leaderboard

Metric: Plan success rate (%; the VLM's high-level Pick/Place/Open/Close plan, predicted from a real human video and an instruction, run through a symbolic executor and checked against the task's success predicate; 155 instances). Source: arxiv.org. Saturation forecast: Around December 2026. 9 models tracked.

Top models

#ModelScore
1GPT-5.460.6
2Gemini 3.1 Pro (Preview)60
3GPT-5.154.6
4Gemini 3 Flash53.5
5Qwen 3 VL 235B A22B (Thinking)31
6Qwen 3 VL 235B A22B Instruct14.8
7Gemini 3.1 Flash Lite14.2

Interactive version: theaggregate.ai/benchmark?slug=watchact-state-change · How It Works · Data refreshed daily, snapshot 2026-09-26.