MEGA-Bench Task - DOCCI Image Description Long: leaderboard
Metric: Task Score (%). Source: huggingface.co. 44 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 2.5 Pro (Preview 03-25) | 86.4 |
| 2 | GPT-4o | 84.3 |
| 3 | InternVL3-78B | 80 |
| 4 | InternVL3-8B | 80 |
| 5 | Claude 3.5 Sonnet (20241022) | 79.3 |
| 6 | Claude 3.5 Sonnet (20240620) | 79.3 |
| 7 | InternVL2.5-78B | 77.9 |
| 8 | Gemini 1.5 Flash (002) | 76.4 |
| 9 | InternVL3-38B | 75.7 |
| 10 | Gemini 1.5 Pro (002) | 75.7 |
| 11 | Qwen 2 VL 72B | 75 |
| 12 | GPT-4o Mini | 72.1 |
| 13 | Gemma 3 27B (IT) | 68.6 |
| 14 | Qwen 2 VL 7B | 67.9 |
| 15 | Llama 4 Scout Base | 67.1 |
Interactive version: theaggregate.ai/benchmark?slug=mega-bench-task-docci-image-description-long · How It Works · Data refreshed daily, snapshot 2026-09-05.