OpenVLM MTVQA - Korean: leaderboard
Metric: Accuracy (%). Source: huggingface.co. 157 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-4.1 Mini | 41.9 |
| 2 | GPT-4.1 (2025-04-14) | 36.4 |
| 3 | Gemini 1.5 Pro | 35.8 |
| 4 | InternVL3-78B | 34.9 |
| 5 | Qwen 2 VL 72B | 33.2 |
| 6 | InternVL3-14B | 32.6 |
| 7 | InternVL3-8B | 31.9 |
| 8 | Claude 3.7 Sonnet | 31.5 |
| 9 | GPT-4.1 Nano | 30.6 |
| 10 | Qwen 2 VL 7B | 30.1 |
| 11 | Gemma 3 27B | 25.6 |
| 12 | Gemini 1.5 Flash | 25.1 |
| 13 | Qwen 2 VL 2B | 23.7 |
| 14 | Gemma 3 12B | 20.1 |
| 15 | Gemma 3 4B | 17 |
Interactive version: theaggregate.ai/benchmark?slug=openvlm-mtvqa-korean · How It Works · Data refreshed daily, snapshot 2026-09-19.