LivingScreen - GUI Action: leaderboard

Metric: Success rate (%) on L1 atomic GUI actions (liking, collecting, commenting, reporting, swiping, seeking), graded on the final platform state; LivingScreen: 499 Chinese short-video tasks in a browser replica of a short-video app (1,528 real videos, about 5.6 per feed); a single-agent loop issues GUI actions plus watch and wait primitives from screenshots or recorded clips, 30-step cap, temperature 0.6, high reasoning effort; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 13 models tracked.

Top models

#ModelScore
1Gemini 3.5 Flash (High)90.4
2Seed 2.0 Pro (High)88.6
3Gemini 3.1 Pro (Preview) (High)87.7
4GLM-5V Turbo68.4
5Claude Sonnet 4.6 (High)67.5
6Claude Opus 4.6 (High)64
7GPT-5.5 (High)50.9
8Kimi K2.5 (Thinking)43.9
9GPT-5.4 (High)37.7

Interactive version: theaggregate.ai/benchmark?slug=livingscreen-gui-action · How It Works · Data refreshed daily, snapshot 2026-09-29.