OCR benchmark
Higher is better
| # | Model | Score | Source |
|---|---|---|---|
| ★ | Qwen2-VL 72B | 87.6 | codesota-api |
| 2 | InternVL2-76B | 87.2 | codesota-api |
| 3 | Gemini 1.5 Pro | 86.5 | codesota-api |
| 4 | PaLI-X 55B | 86.1 | codesota-api |
| 5 | NVLM-D 1.0 72B | 85.4 | codesota-api |
| 6 | NVLM-X 1.0 72B | 85.2 | codesota-api |
| 7 | NVLM-H 1.0 72B | 85.2 | codesota-api |
| 8 | VILA-1.5 40B | 84.3 | codesota-api |
| 9 | LLaVA-NeXT 34B | 83.7 | codesota-api |
| 10 | LLaVA-NeXT 13B | 82.8 | codesota-api |
| 11 | CogVLM-17B | 82.3 | codesota-api |
| 12 | LLaVA-NeXT 7B (Mistral) | 82.2 | codesota-api |
| 13 | BLIP-2 | 82.19 | codesota-api |
| 14 | LLaVA-NeXT 7B (Vicuna) | 81.8 | codesota-api |
| 15 | Pixtral Large | 80.9 | codesota-api |
| 16 | Llama 3-V 405B | 80.2 | codesota-api |
| 17 | LLaVA-1.5 13B | 80 | codesota-api |
| 18 | LLaVA-1.5 | 80 | codesota-api |
| 19 | Llama 3-V 70B | 79.1 | codesota-api |
| 20 | Pixtral-12B | 78.6 | codesota-api |
| 21 | GPT-4o | 78.5 | codesota-api |
| 22 | Llama 3.2 90B Vision Instruct | 78.1 | codesota-api |
| 23 | GPT-4V | 77.2 | codesota-api |