OpenVLM MMT-Bench - Counting By Visual Prompting: leaderboard
Metric: Score (%). Source: huggingface.co. 207 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-4.1 Mini | 65 |
| 2 | Qwen 2 VL 2B | 55 |
| 3 | Claude 3.5 Sonnet | 55 |
| 4 | Grok 2 (1212) | 50 |
| 5 | Qwen 2 VL 72B | 45 |
| 6 | Claude 3.7 Sonnet | 45 |
| 7 | GPT-4.1 (2025-04-14) | 45 |
| 8 | GPT-4.1 Nano | 40 |
| 9 | Gemma 3 4B | 40 |
| 10 | Gemma 3 27B | 35 |
| 11 | Gemma 3 12B | 35 |
| 12 | Claude 3 Sonnet | 35 |
| 13 | InternVL3-78B | 30 |
| 14 | InternVL3-8B | 30 |
| 15 | InternVL3-38B | 30 |
Interactive version: theaggregate.ai/benchmark?slug=openvlm-mmt-bench-counting-by-visual-prompting · How It Works · Data refreshed daily, snapshot 2026-09-05.