EvoGUI-Bench - Inverse Action and Value: leaderboard
Metric: Joint action and value accuracy (%; share of 1,000 consecutive screenshot pairs for which the model predicts both the click, type or select action and its typed or selected value; on EvoGUI-Bench, 3,000 zero-shot visual question answering probes mined from Mind2Web and WebLINX web trajectories across 120 domains, screenshots at max_image_side 512, temperature 0; higher is better). Source: arxiv.org. Saturation forecast: Around 2031. 28 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Flash (Preview) | 65.4 |
| 2 | Qwen 3.5 397B A17B | 56.6 |
| 3 | Qwen 3.6 35B A3B | 56.6 |
| 4 | Qwen 3.5 122B A10B | 56.4 |
| 5 | Doubao-Seed-2.0-Pro-260215 | 55.7 |
| 6 | Qwen 3.6 27B | 55.1 |
| 7 | Qwen 3 VL 235B A22B | 55.1 |
| 8 | Qwen 3.5 35B A3B | 54.8 |
| 9 | Qwen 3.5 27B | 54.8 |
| 10 | MiMo-V2.5 | 54.5 |
| 11 | GLM-5V Turbo | 54 |
| 12 | Kimi K2.5 | 52.8 |
| 13 | Kimi K2.6 | 52.6 |
| 14 | GPT-5 Mini | 52.5 |
| 15 | GLM-4.5V | 52.3 |
Interactive version: theaggregate.ai/benchmark?slug=evogui-bench-inverse-action-and-value · How It Works · Data refreshed daily, snapshot 2026-09-29.