KOFFVQA: leaderboard
Korean Open-ended Free-form VQA benchmark: 82 vision-language models evaluated on object recognition, document/table/chart understanding, OCR, and commonsense reasoning in Korean.
Metric: Overall Score (%). Source: huggingface.co. Saturation forecast: Estimated already saturated. 82 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro (Preview 03-25) | 89.67 | #123 |
| 2 | Gemini 2.5 Flash (Preview 04-17) | 89.53 | #237 |
| 3 | Gemini 2.5 Pro (Preview 05-06) | 88.98 | #147 |
| 4 | O3 (2025-04-16) | 88.29 | #121 |
| 5 | O1 (2024-12-17) | 88.25 | #149 |
| 6 | O4 Mini (2025-04-16) | 87.85 | #179 |
| 7 | Gemini 2.5 Flash (Preview 05-20) | 86.76 | #216 |
| 8 | GPT-4.1 Mini | 85.82 | #355 |
| 9 | GPT-4.1 | 85.71 | #248 |
| 10 | GPT-4.5 (Preview) | 85.31 | #244 |
| 11 | GPT-4o (2024-11-20) | 81.96 | #381 |
| 12 | Claude 3.5 Sonnet (20241022) | 80.47 | #291 |
| 13 | Qwen 2.5 VL 72B Instruct | 80.11 | #377 |
| 14 | Llama 4 Scout Instruct | 79.93 | #539 |
| 15 | Gemini 2.0 Flash (001) | 79.09 | #360 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=koffvqa · How It Works · Data refreshed daily, snapshot 2026-10-20.