OpenVLM CCBench - Historical Figures: leaderboard
CCBench Historical Figures accuracy from the OpenCompass OpenVLM Leaderboard (VLMEvalKit): multiple-choice questions on Chinese culture about historical figures. The category holds 35 of the 510 questions and is scored with CircularEval (correct under every rotation of the options). Accuracy in percent, higher is better.
Metric: Accuracy (%). Source: huggingface.co. Saturation forecast: Estimated already saturated. 228 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 2 VL 72B | 60 | #450 |
| 2 | InternVL3-78B | 54.3 | #361 |
| 3 | GPT-4.1 (2025-04-14) | 45.7 | #257 |
| 4 | GPT-4.1 Mini | 42.9 | #363 |
| 5 | InternVL3-14B | 37.1 | #506 |
| 6 | Qwen 2 VL 7B | 28.6 | #743 |
| 7 | Gemma 3 27B | 28.6 | #606 |
| 8 | Qwen-VL-Max | 28.6 | #475 |
| 9 | InternVL3-38B | 25.7 | #413 |
| 10 | MiniCPM-V-2.6 | 25.7 | #827 |
| 11 | Qwen 2.5 VL 7B | 25.7 | #641 |
| 12 | Claude 3.5 Sonnet (20241022) | 22.9 | #302 |
| 13 | Gemma 3 12B | 20 | #674 |
| 14 | Grok 2 (1212) | 20 | #502 |
| 15 | Moonshot v1 8K | 20 | #622 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=openvlm-ccbench-historical-figures · How It Works · Data refreshed daily, snapshot 2026-11-05.