CulMind: leaderboard
Evaluating Multimodal Large Language Models (MLLMs) in Chinese Cultural Heritage (CCH) requires fine-grained reasoning over visual, textual, stylistic, and historical clues. However, existing CCH benchmarks mainly emphasize final-answer accuracy, while the accuracy and completeness of reasoning processes remain underexplored. To address this gap, we introduce CulMind and CulMind-R: a high-quality
Metric: S (self-reported). Source: benchmarklist.com. Saturation forecast: Around September 2027. 14 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Flash (Preview) | 50.7 | #82 |
| 2 | Qwen 3 VL 8B Instruct | 48.7 | #420 |
| 3 | Qwen 3 VL 32B Instruct | 46.6 | #283 |
| 4 | GPT-5.5 | 45.9 | #31 |
| 5 | Gemini 2.5 Flash | 44.7 | #245 |
| 6 | GLM-4.5V | 43.8 | #346 |
| 7 | InternVL3-8B | 42.8 | #620 |
| 8 | InternVL3-14B | 42.5 | #505 |
| 9 | GPT-5.4 Mini | 39.9 | #212 |
| 10 | Qwen 3.5 27B | 39.7 | #187 |
| 11 | Qwen 3.5 9B | 37.5 | #379 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=culmind · How It Works · Data refreshed daily, snapshot 2026-10-20.