LM Studio Panel
Your browser straight to your on-device LM Studio (v1 REST API) — manage models, test chats
Connect LM Studio
Default address http://localhost:1234 (LM Studio 0.4.0+). This page talks from your browser straight to your on-device service — no third party ever sees the traffic. Connection details (incl. API token) are AES-GCM-encrypted and kept in your own browser only; wipe them anytime in one click.
What is this?
LM Studio is the common way to run large models locally, and it ships a local API server (default http://localhost:1234). Since 0.4.0 it offers a proper v1 REST API covering models, load/unload, downloads and chat. This page is that API's browser control panel: enter the server address (plus an API token if you use one), hit "Connect" to see downloaded models, expand any model to tune context length, Flash Attention and more into memory — or unload to free it. The chat test bench streams, takes System Prompts, temperature and the usual knobs, with multi-turn memory carried by the server's response_id mechanism — no pasting history back and forth.
How it relates to LM Studio's own UI and OpenAI-compatible endpoints
The LM Studio app already does all this — this panel's value is a lightweight browser-side entry point: per-request stats while debugging API integrations (tokens/s, time to first token, token usage), and one-handed management of your own service from a tablet or phone browser (over a LAN address). If you call it from code, LM Studio also serves OpenAI / Anthropic-compatible endpoints — no conflict with this panel.
What to know about browser-to-localhost
Requests from an https page to localhost count as "public page accessing the local network", which a few browser versions block. If connecting fails, walk the three self-checks in the FAQ above; request errors also distinguish "server not started / wrong port / blocked by browser / bad token / version too old" — follow the hint.