Add Support for Multiple OCR/LLM Models (Beyond OpenAI) #4
Description
Description:
Currently, the OCR service relies exclusively on a single model provider: OpenAI. To improve flexibility, we should extend support to integrate with other open-source or self-hostable models.
✅ Goals
- Add support for open-source LLMs like:
llama.cpp-based models- Mistral and Anthropic models
- Support Hugging Face-hosted models or local inference with
transformers/vLLM - Add a configuration switch to select model backend (
openai,local,hf,anthropic, etc.)
Priority: Medium to High