MEGA-Bench Task - Vibe Eval: leaderboard
Metric: Task Score (%). Source: huggingface.co. 27 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 2 VL 72B | 69.3 |
| 2 | Gemini 1.5 Pro (002) | 69.3 |
| 3 | GPT-4o | 66.4 |
| 4 | Gemini 1.5 Flash (002) | 60.7 |
| 5 | Claude 3.5 Sonnet (20241022) | 60 |
| 6 | GPT-4o Mini | 57.1 |
| 7 | Qwen 2 VL 7B | 55.7 |
| 8 | InternVL2.5-78B | 55 |
| 9 | Pixtral-12B | 51.4 |
| 10 | Claude 3.5 Sonnet (20240620) | 49.3 |
| 11 | Qwen 2 VL 2B | 47.9 |
| 12 | InternVL2-8B | 40.7 |
| 13 | InternVL2.5-2B | 35 |
| 14 | Aquila-VL-2B | 20 |
Interactive version: theaggregate.ai/benchmark?slug=mega-bench-task-vibe-eval · How It Works · Data refreshed daily, snapshot 2026-09-05.