Feature Request: Support for Local LLMs (LM Studio) #2
Description
Summary
It would be great to add support for using local language models through LM Studio.
Motivation
Many users prefer running models locally instead of relying on external APIs. This can help with:
- Privacy (no data sent to third-party services)
- Offline usage
- Avoiding API costs
- Greater control over models and configurations
LM Studio is a popular tool that makes it easy to run and manage local models, so supporting it would make LLMMap more flexible and accessible.
Proposed Idea
Add an option to use LM Studio as a backend for running queries. Since LM Studio exposes an API similar to OpenAI, it may integrate smoothly with the existing structure.
A possible approach could be to introduce a dedicated loader (e.g., LLM_LMStudio) that:
- Connects to LM Studio’s local server (typically
http://localhost:1234/v1) - Uses the OpenAI-compatible API endpoints exposed by LM Studio
- Supports both chat and completion-style requests
Ideally, users could:
- Select LM Studio as a backend
- Configure a local endpoint
- Run queries the same way they would with other providers
Benefits
- Expands support for local-first workflows
- Useful for research, lab environments, and privacy-focused users
- Reduces dependency on external services
Additional Notes
Happy to help test or contribute if this is something you'd consider adding.