KOFFVQA - Korean OCR: leaderboard
KOFFVQA subscore for reading and using Korean text visible in images.
Metric: Score (%). Source: huggingface.co. Saturation forecast: Estimated already saturated. 82 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4o Mini (2024-07-18) | 100 | #558 |
| 2 | GPT-4.1 | 95 | #248 |
| 3 | Gemini 2.5 Pro (Preview 03-25) | 95 | #123 |
| 4 | O4 Mini (2025-04-16) | 95 | #179 |
| 5 | O3 (2025-04-16) | 95 | #121 |
| 6 | Qwen 2.5 VL 72B Instruct | 95 | #377 |
| 7 | O1 (2024-12-17) | 95 | #149 |
| 8 | Gemini 2.5 Pro (Preview 05-06) | 95 | #147 |
| 9 | GPT-4.5 (Preview) | 95 | #244 |
| 10 | Gemini 2.0 Flash (001) | 93.5 | #360 |
| 11 | GPT-4o (2024-11-20) | 91.5 | #381 |
| 12 | GPT-4.1 Mini | 90 | #355 |
| 13 | Gemini 2.0 Flash (Preview) | 90 | #370 |
| 14 | Qwen2.5-Omni-7B | 90 | #624 |
| 15 | Llama 4 Scout Instruct | 88.5 | #539 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=koffvqa-korean-ocr · How It Works · Data refreshed daily, snapshot 2026-10-20.