Feature Proposal: Shift to Local LLM-based Translation & OCR as the Core Engine

[RichardDBrown]

Hi everyone! 👋

First of all, I really love this project and appreciate all the work that has gone into it.

I'd like to share a suggestion for the future direction of Pot:

The Idea

With the rapid development of AI and large language models (LLMs), I believe it's time to consider a significant shift in architecture:

1. Replace traditional translation services with LLMs – Local LLMs have become incredibly capable for translation tasks, often matching or exceeding the quality of traditional API-based services.

2. Prioritize local/offline LLM support – Many open-source models (like Ollama-compatible models) now run efficiently on consumer hardware and provide excellent translation and OCR results. This would give users:

   • Better privacy (no data sent to external servers)
   • No API costs or rate limits
   • Offline functionality

3. Deprecate legacy translation plugins – As local LLMs become the standard, the older translation service integrations could be phased out or made optional.

Why This Matters

The AI landscape has changed dramatically. Local models can now handle both translation and image/text recognition (OCR) with impressive accuracy. I think embracing this shift could define the next evolution of Pot.

I'd love to hear the community's thoughts on this direction! 🚀