Use qwen2.5vl:7b (Ollama) to ocr image not work

[hhoangcvp]

Hello, I am currently using Docker Compose to build the model.
However, when I use qwen2.5vl:7b to perform OCR on an image via Web LobeChat, the system does not return the expected answer.
But when I call the API directly via Ollama, it works fine. Is there any way to solve this issue?
Request: Please help me OCR the following image

• Gemini + GPT => works fine
• qwen2.5vl:7b (via Web LobeChat) => not working
• qwen2.5vl:7b (API call via Postman to Ollama) => works fine
  Is there any configuration or solution to make (OCR Image) the qwen2.5vl:7b model work properly on Web LobeChat?