MEGA-Bench Task - GUI Chat Hard: leaderboard
Metric: Task Score (%). Source: huggingface.co. 44 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 2 VL 72B | 61 |
| 2 | InternVL2-8B | 60.3 |
| 3 | InternVL3-8B | 60 |
| 4 | Gemini 2.5 Pro (Preview 03-25) | 58.8 |
| 5 | InternVL3-38B | 56.5 |
| 6 | InternVL3-78B | 54.2 |
| 7 | Gemma 3 27B (IT) | 51.2 |
| 8 | Llama 4 Scout Base | 50.4 |
| 9 | GPT-4o | 49.4 |
| 10 | InternVL2.5-2B | 48.1 |
| 11 | Claude 3.5 Sonnet (20241022) | 45.8 |
| 12 | InternVL2.5-78B | 45.8 |
| 13 | Claude 3.5 Sonnet (20240620) | 45.8 |
| 14 | Qwen 2 VL 7B | 45.5 |
| 15 | Qwen 2 VL 2B | 45.5 |
Interactive version: theaggregate.ai/benchmark?slug=mega-bench-task-gui-chat-hard · How It Works · Data refreshed daily, snapshot 2026-09-05.