---
title: "LM Studio: what it connects to on a Mac, its local server port, and whether to block it · Kilobat"
url: "https://kilobat.app/processes/lm-studio"
description: "LM Studio runs language models on your Mac. When it goes online, which port its local server uses, how to share it on a network, and what breaks if you block it."
---

# What is LM Studio?

LM Studio runs language models on your own Mac. Its docs list exactly which actions need the internet, and say that chatting with a downloaded model does not.

**Fix this with your agent**Copies a prompt for Claude Code, Codex or any agent that can use Terminal.

I want to know what LM Studio is doing on my Mac and whether its network use is normal. Read https://kilobat.app/processes/lm-studio.md and work through it with me. Install the Kilobat CLI if it’s missing: brew install seifip/kilobat/kilobat. Run the guide’s Terminal and kilobat checks yourself and tell me what you find. List the steps only I can do. Ask before you change any setting.

Process: `LM Studio` · Also seen as: `lms`, `llmster` · Made by LM Studio · Applies to: LM Studio on macOS 14 Sonoma and later. Checked on 8 November 2026 against LM Studio's documentation. · Checked: 8 November 2026

## Short answer

LM Studio runs models locally. It goes online to search and download models and runtimes and to check for updates. Its API server is local unless you share it.

Safe to block

Blocking LM Studio's internet access only stops search, downloads and update checks. Chat with downloaded models keeps working.

Path

`/Applications/LM Studio.app`

## What it is

LM Studio's offline-operation page says that the app does not need the internet for its core functions. These are chatting with models, chatting with documents and running a local server.

The page says that nothing you type into a chat with a downloaded model leaves your device. It says the same about documents that you drop in for chat.

It lists what does need a connection. These are searching for models, downloading models, the Discover catalogue, downloading runtimes, and checking for app updates.

For model search and catalogue statistics it gives `huggingface.co` as an example host. It says the in-app updater makes a request when the app opens, to check for a new version.

The server docs say that you start the API server from the Developer tab or with `lms server start`. The examples in the OpenAI-compatible docs use port `1234`.

By default the server stays on localhost. Server Settings has a port field, a Require Authentication switch and a Serve on Local Network switch.

## Why it uses the network

- **Model downloads.** LM Studio says that downloading a model needs a stable and decently fast connection. You can also sideload models that you got elsewhere.
- **Discover and search.** LM Studio says that each version ships with a built-in catalogue. The Discover tab asks the network for model stats and download options.
- **Runtimes.** LM Studio says that runtimes are packaged engines, such as llama.cpp. The app contacts the network to check for and download them.
- **Update checks.** LM Studio says that the updater asks for a new version when the app opens. It shows a notification if there is one.
- **Local API server.** Apps on this Mac, and on your network if you allow it, send requests to the server. These stay local, according to LM Studio.

## How much is normal

With a model already downloaded, LM Studio's docs say that chat and server use need no internet. Expect little internet traffic then.

A short request at launch is expected, because LM Studio says that the updater checks for a new version.

Connections to the LM Studio server on port 1234 come from your own tools. They stay on this Mac or your network.

## When it uses a lot, and why

- **Downloading a model.** Model files are large. A download uses as much of your connection as it can.
- **Downloading a runtime.** LM Studio says that runtimes are separate packages, which you can download without a full app update.
- **Serving other devices.** With Serve on Local Network on, other devices send prompts to this Mac. LM Studio says that any bind other than 127.0.0.1 exposes the server beyond localhost.

## Is it safe to block?

Safe to block

Blocking LM Studio's internet access only stops search, downloads and update checks. Chat with downloaded models keeps working.

What breaks if you block it

- The Discover tab cannot search or show model stats.
- Models and runtimes cannot download.
- The in-app updater cannot check for a new version.
- If you block the local network, other devices cannot reach the API server.

Blocking means a firewall rule that refuses its connections. Don’t quit or disable a macOS process to save data: macOS starts it again, and other features depend on it.

## Reduce its use

1. To keep the API server on this Mac, leave Serve on Local Network off in the Developer tab's Server Settings.
2. If you share the server, turn on Require Authentication. LM Studio recommends authentication for any bind other than localhost.
3. Download models on a network with room for them. You can also sideload models that you already have.

## See how much it uses

1. Open Activity Monitor and search for `LM Studio`. The Network tab shows the bytes that the app sent and received.

## With Kilobat

1. In the Kilobat app list, search for `LM Studio`. Its row shows the live rate and the hosts that it talks to.
2. Ask the same question in Terminal.

   ```
   kilobat explain "LM Studio"
   ```

The live view and the explanations are free. History in the app needs Kilobat Pro. Blocking needs Pro and the network filter: see the [network filter](https://kilobat.app/docs/blocking#filtering), [Protection](https://kilobat.app/docs/blocking#protection) and the [hotspot data saver](https://kilobat.app/save-hotspot-data).

## What isn’t documented

LM Studio's pages do not list the hosts for the updater, runtimes or catalogue. They give `huggingface.co` only as an example for search and catalogue requests.

The pages do not name the processes that the app shows in Activity Monitor. The app path here is the usual install name, not a path from LM Studio.

The pages do not say whether the app sends usage data. They say that your chats and documents stay on your device.

## Questions

**Does LM Studio work offline?**

Yes. LM Studio says that chatting with downloaded models, chatting with documents and running a local server need no internet.

**Which port does the LM Studio server use?**

The OpenAI-compatible examples assume port 1234. Server Settings has a field to change it.

**How do I share the server on my network?**

Turn on Serve on Local Network in Server Settings, or run `lms server start --bind 0.0.0.0`. LM Studio recommends turning on authentication first.

## Sources

- [LM Studio Docs: Offline Operation](https://lmstudio.ai/docs/app/offline)
- [LM Studio Docs: Server Settings](https://lmstudio.ai/docs/developer/core/server/settings)
- [LM Studio Docs: Serve on Local Network](https://lmstudio.ai/docs/developer/core/server/serve-on-network)

Checked on 8 November 2026.

## Related

- [ollama](https://kilobat.app/processes/ollama): Ollama runs language models locally. It listens on 127.0.0.1 port 11434 by default and downloads models from the internet when you pull them.
