MEGA-Bench Task - Counting — leaderboard
Metric: Task Score (%). Source: huggingface.co. 27 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude 3.5 Sonnet (20241022) | 85.7 |
| 2 | Qwen 2 VL 72B | 78.6 |
| 3 | InternVL2.5-78B | 78.6 |
| 4 | Gemini 1.5 Pro (002) | 71.4 |
| 5 | Claude 3.5 Sonnet (20240620) | 71.4 |
| 6 | GPT-4o | 64.3 |
| 7 | InternVL2-8B | 50 |
| 8 | Qwen 2 VL 7B | 42.9 |
| 9 | Pixtral-12B | 42.9 |
| 10 | Gemini 1.5 Flash (002) | 42.9 |
| 11 | Qwen 2 VL 2B | 35.7 |
| 12 | GPT-4o Mini | 35.7 |
| 13 | Aquila-VL-2B | 28.6 |
| 14 | MiniCPM-V-2.6 | 21.4 |
| 15 | InternVL2.5-2B | 14.3 |
Interactive version: theaggregate.ai/benchmark?slug=mega-bench-task-counting · How the rankings work · Data refreshed daily, snapshot 2026-07-22.