WatchAct - Moment: 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; 135 instances). Source: arxiv.org. Saturation forecast: Around September 2028. 9 models tracked.

Top models

#ModelScore
1Gemini 3 Flash23.1
2Gemini 3.1 Pro (Preview)22.7
3Qwen 3 VL 235B A22B (Thinking)12.1
4GPT-5.17.4
5Qwen 3 VL 235B A22B Instruct5.9
6Gemini 3.1 Flash Lite4.6
7GPT-5.44.5

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