LM Studio

LM Studio is a desktop app for running models locally. Its built-in server exposes an OpenAI-compatible API at http://localhost:1234/v1, which BitRouter fronts as one provider block.

Prerequisites

• BitRouter installed, with a bitrouter.yaml (scaffold one with bitrouter init).

• LM Studio's server running with a model loaded — from the app's Developer tab, or the CLI:

  lms server start                       # default port 1234
  lms load llama-3.1-8b-instruct

Add LM Studio to BitRouter

# bitrouter.yaml
providers:
  lmstudio:
    api_base: http://localhost:1234/v1
    api_protocol:
      - "*": chat_completions
    models:
      - id: llama-3.1-8b-instruct

The models id is the model identifier LM Studio shows for the loaded model. Confirm it with lms ps or by hitting http://localhost:1234/v1/models.

Route to it

bitrouter route lmstudio:llama-3.1-8b-instruct

Then start BitRouter and send a request.

Learn more

• LM Studio — OpenAI-compatible API
• Model fallback — fail over from LM Studio to a hosted model.