SideNote ProUse LM Studio with SideNote Pro
LM Studio is a desktop application for downloading and running local models, with a built-in server that speaks the OpenAI-compatible API. If you prefer a graphical model manager to a command line, it is the easier local route.
Setting it up
- Install LM Studio and download a modelLM Studio's own interface handles browsing and downloading. Pick something sized for your hardware; its interface will tell you what fits.
- Start the local serverThis is the step people miss. Downloading a model is not the same as serving it. LM Studio has an explicit local server view, and it must be running with a model loaded.
- Add an LM Studio profile in SideNote ProSettings, AI providers, add from the LM Studio preset.
- Test and choose the modelThe connection test reports what the server is offering. Select the loaded model from the picker, or type its identifier.
| Field | Value |
|---|---|
| Provider name | LM Studio |
| Base URL | http://localhost:1234/v1 |
| Model | The identifier of the model you loaded |
| API key | Empty for a default local server |
| Reasoning effort | Provider default |
If the connection test fails, work through it in this order: is the LM Studio server actually started, is a model loaded, and is the port still 1234. Changing the port in LM Studio means changing the base URL in SideNote Pro to match.
How this differs from Ollama
For SideNote Pro's purposes, hardly at all. Both are local OpenAI-compatible servers, and everything in the Ollama guide about expectations, speed, image support and context size applies here unchanged.
The difference is in the experience of managing models. LM Studio is graphical, with model browsing, hardware compatibility hints and an explicit server toggle. Ollama is command-line first and lighter to run in the background. Which one suits you is a matter of taste, not of what SideNote Pro can do.
The same caveats apply
- Model capability is bounded by your hardware, not by SideNote Pro.
- Image input depends on the loaded model supporting it.
reasoning_effortmay be ignored. Leave it on *Provider default*.- Usage metadata may be absent, in which case local token estimates are used.
- The server must be running for the profile to work. A local profile fails differently from a hosted one, and *unreachable endpoint* almost always means the server stopped.
SideNote Pro does not supply or guarantee third-party local model software. It sends requests to the address you configured.