OpenVLM MMT-Bench - Counting By Reasoning: leaderboard
Metric: Score (%). Source: huggingface.co. 207 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude 3.5 Sonnet | 95 |
| 2 | Qwen 2 VL 72B | 90 |
| 3 | Pixtral-12B | 90 |
| 4 | Claude 3.7 Sonnet | 90 |
| 5 | Grok 2 (1212) | 90 |
| 6 | GPT-4.1 (2025-04-14) | 90 |
| 7 | Llama 3.2 90B Vision Instruct | 90 |
| 8 | Claude 3 Sonnet | 85 |
| 9 | InternVL3-8B | 80 |
| 10 | InternVL3-38B | 80 |
| 11 | Llama 3.2 11B Instruct | 80 |
| 12 | Gemini 1.5 Pro | 80 |
| 13 | Gemini 1.5 Flash | 80 |
| 14 | GPT-4.1 Mini | 75 |
| 15 | InternVL3-78B | 75 |
Interactive version: theaggregate.ai/benchmark?slug=openvlm-mmt-bench-counting-by-reasoning · How It Works · Data refreshed daily, snapshot 2026-09-05.