MEGA-Bench Task - Web Action Grounding — leaderboard
Metric: Task Score (%). Source: huggingface.co. 44 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemma 3 27B (IT) | 85.7 |
| 2 | Gemini 2.5 Pro | 78.6 |
| 3 | InternVL3-78B | 78.6 |
| 4 | Qwen 2 VL 72B | 78.6 |
| 5 | Qwen 2 VL 7B | 78.6 |
| 6 | Claude 3.5 Sonnet (20241022) | 78.6 |
| 7 | Claude 3.5 Sonnet (20240620) | 78.6 |
| 8 | InternVL3-38B | 71.4 |
| 9 | GPT-4o | 64.3 |
| 10 | GPT-4o Mini | 64.3 |
| 11 | Gemini 1.5 Pro (002) | 64.3 |
| 12 | Llama 4 Scout Base | 64.3 |
| 13 | Gemini 2.0 Flash (Preview) | 64.3 |
| 14 | InternVL3-14B | 57.1 |
| 15 | Ovis2-8B | 57.1 |
Interactive version: theaggregate.ai/benchmark?slug=mega-bench-task-web-action-grounding · How the rankings work · Data refreshed daily, snapshot 2026-07-22.