MEGA-Bench Task - Video Short Title: leaderboard
Metric: Task Score (%). Source: huggingface.co. 44 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude 3.5 Sonnet (20241022) | 81.4 |
| 2 | Claude 3.5 Sonnet (20240620) | 75 |
| 3 | Gemini 2.5 Pro (Preview 03-25) | 74.3 |
| 4 | GPT-4o | 73.6 |
| 5 | Gemma 3 12B (IT) | 70.7 |
| 6 | InternVL3-38B | 69.3 |
| 7 | Gemma 3 27B (IT) | 67.9 |
| 8 | InternVL3-78B | 67.9 |
| 9 | InternVL2.5-78B | 67.1 |
| 10 | Gemini 1.5 Flash (002) | 66.4 |
| 11 | Qwen 2 VL 72B | 65.7 |
| 12 | Pixtral-12B | 65 |
| 13 | GPT-4o Mini | 64.3 |
| 14 | Qwen 2 VL 7B | 63.6 |
| 15 | Gemini 1.5 Pro (002) | 63.6 |
Interactive version: theaggregate.ai/benchmark?slug=mega-bench-task-video-short-title · How It Works · Data refreshed daily, snapshot 2026-09-05.