OpenVLM CCBench - Historical Figures: leaderboard
Metric: Accuracy (%). Source: huggingface.co. 228 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 2 VL 72B | 60 |
| 2 | InternVL3-78B | 54.3 |
| 3 | GPT-4.1 (2025-04-14) | 45.7 |
| 4 | GPT-4.1 Mini | 42.9 |
| 5 | InternVL3-14B | 37.1 |
| 6 | Qwen 2 VL 7B | 28.6 |
| 7 | Gemma 3 27B | 28.6 |
| 8 | Claude 3.5 Sonnet (20241022) | 22.9 |
| 9 | Grok 2 (1212) | 20 |
| 10 | Gemma 3 12B | 20 |
| 11 | Moonshot v1 8K | 20 |
| 12 | Llama 3.2 90B Vision Instruct | 20 |
| 13 | InternVL2-8B | 17.1 |
| 14 | Qwen 2 VL 2B | 17.1 |
| 15 | Aquila-VL-2B | 17.1 |
Interactive version: theaggregate.ai/benchmark?slug=openvlm-ccbench-historical-figures · How It Works · Data refreshed daily, snapshot 2026-09-19.