How to Chat with LM Studio or Ollama From Anywhere Using Android and Tailscale

Published on June 14, 2026 • By LMSA
How to Chat with LM Studio or Ollama From Anywhere Using Android and Tailscale

If you've gotten comfortable running models locally with LM Studio or Ollama, you've probably hit the same wall everyone does eventually: it's all stuck on one machine. You load up a model, chat with it from your desktop, and the second you walk away with your phone, that whole setup might as well not exist. The good news is that fixing this doesn't require renting cloud GPUs or exposing your computer to the open internet. With a free Android app called LMSA and a free networking tool called Tailscale, you can turn your home PC into a private AI server that follows you everywhere — coffee shop, work, another country, doesn't matter.

This guide walks through the whole process from scratch, including creating your Tailscale account, which trips people up more often than you'd expect. No prior networking knowledge needed. By the end, you'll be able to open LMSA on your phone, tap into the model running on your PC, and chat with it as if it were sitting in your pocket.

What You're Actually Building

Before diving in, it helps to understand the shape of what you're setting up, because it makes troubleshooting much easier later.

LM Studio and Ollama both run a small local server on your computer. By default, that server only listens to your computer itself — it's deliberately closed off, even from other devices on your own Wi-Fi. The first part of this guide opens that server up so devices on your network can talk to it.

Tailscale is the second piece. It works on Windows, macOS, Linux, iOS, and Android, and once you install it on two or more devices and sign into the same account, those devices form what's called a "tailnet" — essentially a private, encrypted network that exists no matter where each device physically is. Every device added to a tailnet gets a unique name and a private IP address, so your phone can reach your PC using an address that works identically whether you're both on the same couch or 3,000 miles apart.

LMSA is the final piece — the app on your phone that actually displays the chat interface and sends your messages to whichever model is loaded on your PC.

Part 1: Creating Your Tailscale Account

This is the step that's easy to assume is obvious, but it's worth doing carefully, because the account you create here is the one every device you connect later has to use.

  1. On your computer, go to tailscale.com and click Get Started (or Sign Up) in the top right corner.
  2. You'll land on a page titled "Sign up with your identity provider." Tailscale doesn't use its own usernames and passwords — instead, it piggybacks on an account you already have. You'll see options like Google, Microsoft, GitHub, Apple, or Passkey.
  3. Pick whichever one you're most comfortable with. For most home users, Google is the simplest choice since almost everyone already has one. Click it.
  4. Log in with that account's normal credentials. If you're already signed into that account in your browser, this may happen instantly with no extra typing.
  5. After authenticating, Tailscale automatically creates your tailnet — your own private network — and drops you into the admin console, a web dashboard at login.tailscale.com/admin. This is where you'll later see every device you connect.

A couple of things worth knowing at this stage:

  • If you signed up with a personal email like a Gmail address, you're automatically placed on the Personal plan, which is free and includes everything this guide needs — generous device limits, encrypted connections, and the admin console.
  • There's no separate "create account" step beyond this. The act of signing in with your chosen provider for the first time is the account creation. Just remember which provider and which email you used, because every other device you add — your PC, your phone — needs to sign in with that exact same one.
  • You don't need to install anything yet to complete this step. The account exists the moment you've authenticated through the browser.

With the account created, you're ready to bring devices into it.

Part 2: Setting Up Your PC's AI Server

Option A: If You're Using LM Studio

  1. Open LM Studio on your computer and load the model you want to chat with.
  2. Click the Developer tab in the left sidebar — it looks like a </> icon.
  3. Find the toggle labeled "Serve on Local Network" and switch it on. Enabling this allows the API server running on your machine to be accessible by other devices, rather than only the machine itself.
  4. Start the local server if it isn't running already.
  5. Once it's running, LM Studio will display a network address, something that looks like http://192.168.1.50:1234. Jot this down — you won't actually need the local IP once Tailscale is set up, but it confirms the server is broadcasting correctly.

A quick note on the newer "LM Link" feature you may have seen mentioned: this is LM Studio's own built-in alternative to the LMSA + Tailscale combo, and it's still in preview. It's worth knowing about, but this guide focuses on the more battle-tested route of pairing LMSA with your own Tailscale network, which gives you more control and works with both LM Studio and Ollama using the same setup.

Option B: If You're Using Ollama

Ollama is more of a background service, so opening it up involves setting one environment variable rather than flipping a switch in a GUI.

By default, Ollama only listens on 127.0.0.1, meaning it talks only to itself. You need to change that to 0.0.0.0, which tells it to accept connections from any device that can reach your computer.

On Windows:

  1. Search for "Environment Variables" in the Start menu and open "Edit the system environment variables."
  2. Click "Environment Variables" near the bottom of the window.
  3. Under "System variables," click "New."
  4. Set the variable name to OLLAMA_HOST and the value to 0.0.0.0.
  5. Click OK on all the open windows, then restart Ollama (quit it from the system tray and relaunch, or restart your computer).

On Mac or Linux: You can launch Ollama with this variable set directly from a terminal:

OLLAMA_HOST=0.0.0.0 ollama serve

If you want this to persist permanently rather than just for one session, add the export line to your shell profile (.zshrc, .bashrc, etc.) or configure it as a launch agent / systemd environment variable, depending on your OS.

Once this is set, Ollama listens on port 11434 and is reachable from other devices, including, eventually, your phone over Tailscale.

Part 3: Installing Tailscale on Your PC

  1. Go to tailscale.com/download in your browser and download the installer for your operating system.
  2. Run the installer. On Windows this is a straightforward MSI-based setup; on Mac, you'll get a standard app you drag into Applications; on Linux, most distributions support a one-line install script provided on Tailscale's download page.
  3. When the app launches, it'll prompt you to log in. Choose the same identity provider and account you used in Part 1 — this is the step where mismatched accounts most commonly cause headaches, so double-check you're using the same Google (or other) account.
  4. Once signed in, Tailscale runs quietly in the background (system tray on Windows, menu bar on Mac). You should see a small Tailscale icon confirming it's connected.
  5. After signing in, the device appears in your Tailscale admin console, where every device gets a unique name generated from its hostname. You can rename it to something memorable, like "home-pc" or "desktop," which will make the next steps much easier to follow.

At this point, your computer has joined your private tailnet and has been assigned a Tailscale IP address, typically something starting with 100.x.x.x.

Part 4: Installing Tailscale on Your Android Phone

  1. Open the Google Play Store and search for "Tailscale," or follow the link from the Tailscale download page.
  2. Install the app like any other.
  3. Open it and sign in using the exact same account you used on your PC and created in Part 1. This is the part people most often get wrong — if you sign in with a different Google account or a different SSO provider, your phone and PC will end up on two separate, unrelated tailnets that can't see each other at all.
  4. Once signed in, toggle the connection on. You'll likely get an Android prompt asking permission to set up a VPN connection — this is expected and normal; Tailscale works by creating a lightweight VPN-style tunnel on your device. Approve it.
  5. After a few seconds, your phone should show as "Connected." If you open the admin console in a browser (either on your PC or phone), you should now see both your computer and your phone listed as devices on the same tailnet.

Part 5: Finding Your PC's Tailscale Address

Now you need the Tailscale IP address of your PC — this is the address your phone will use to reach it, and it works the same whether you're on the same Wi-Fi or on opposite sides of the planet.

On Windows or Mac:

  • Click the Tailscale icon in your system tray or menu bar. It usually shows your device's Tailscale IP directly in the dropdown menu.

Using the admin console (works from any device):

  • Open login.tailscale.com/admin/machines in a browser.
  • Find your PC in the device list and note the IP address shown next to it — it'll look like 100.101.102.103.

You can also get this from a terminal/command prompt by running:

tailscale ip -4

Write this address down. It's the key piece of information LMSA needs.

Part 6: Setting Up LMSA on Your Phone

  1. Open the Google Play Store and search for "LMSA" — the full listing is titled "LMSA for LM Studio & Ollama." Install it.
  2. Open the app. You'll land on the main chat interface, but first you need to point it at your server.
  3. Tap on the 'PROVIDER' tab at the top of the screen and select the 'Local Server' option.
  4. Select either LM Studio or Ollama for the connection type.

Here's the part that makes this whole setup work: instead of typing in your PC's regular home network IP (the 192.168.x.x one), you type in the Tailscale IP you found in Part 5.

  • For LM Studio: Enter the address as http://100.101.102.103:1234 (replace with your actual Tailscale IP, keeping the port LM Studio showed you, typically 1234).
  • For Ollama: Enter http://100.101.102.103:11434 (port 11434 is Ollama's default).
  1. Save the settings by tapping the green 'Apply' button in the LMSA app. LMSA should attempt to connect and, assuming both devices are online and connected to Tailscale, pull up the list of models currently loaded on your PC by tapping the 'MODELS' tab at the top of the LMSA app.
  2. Select your model from the list, and you're ready to chat.

Part 7: Testing It From Outside Your Home

This is the moment of truth. Once everything above is configured:

  1. Turn off Wi-Fi on your phone and switch to mobile data, or take your phone to a different location entirely.
  2. Make sure Tailscale is still toggled on in the Tailscale app (it should stay connected in the background automatically).
  3. Open LMSA and send a test message.

If your PC is awake, LM Studio or Ollama is running, and Tailscale is connected on both ends, the message should go through and you'll get a response streamed back, just like you would sitting at your desk.

If it doesn't work right away, here are the most common culprits:

  • Your PC went to sleep. Local AI servers can't respond if the computer is asleep or shut down. Adjust your power settings so the machine stays awake, or at least doesn't fully sleep, while you're away.
  • The model got unloaded. Some setups (Ollama especially) unload models from memory after a period of inactivity to save resources. You can set the OLLAMA_KEEP_ALIVE environment variable to control how long a model stays loaded if this becomes a recurring annoyance.
  • Tailscale isn't actually connected on one end. Double-check both devices show as "Connected" with green status in their respective Tailscale apps.
  • You typed the wrong IP or port. Re-verify the Tailscale IP hasn't changed (it can occasionally, though MagicDNS names are more stable if enabled) and that the port matches what LM Studio or Ollama is actually using.
  • Two different Tailscale accounts. If your phone and PC don't show up on the same admin console device list, you signed into different accounts somewhere along the way. Sign out on one device and sign back in with the correct account from Part 1.

A Word on Security

It's worth understanding why this setup is safe, because the idea of "opening up" your AI server can sound alarming if you're used to hearing warnings about exposing services to the internet.

The key distinction is that you are not exposing anything to the public internet. Tailscale leverages a mesh VPN that keeps your devices connected without exposing them to the public internet, with your chats remaining local while the heavy processing happens on your remote hardware. Nothing about your LM Studio or Ollama server is reachable by anyone except devices you've personally authorized into your tailnet. There's no port forwarding involved, no public IP exposure, and no firewall rules to punch holes in. As far as the broader internet is concerned, your home PC's AI server doesn't exist — it's only visible inside the private network you've built.

The "Serve on Local Network" setting in LM Studio and the OLLAMA_HOST=0.0.0.0 change in Ollama only open the server up to your local network by themselves. Avoiding port forwarding is important — opening ports on a router to expose Ollama or LM Studio directly to the internet can leave a computer vulnerable to scanning and attacks. Tailscale sidesteps all of that by handling the remote connection through its own encrypted tunnels instead.

One more practical point on the LMSA side: all messages are stored locally on the device, and the local chat database is fully encrypted on-device, so even your conversation history isn't sitting on someone else's server somewhere.

You're ready to chat with local AI models anywhere on the go!

Once this is set up, it tends to fade into the background — you stop thinking about "connecting" to your home PC and it just becomes the way you use your local models. The combination covers both major local AI platforms, costs nothing beyond what you've already invested in your hardware, and doesn't require you to trust your conversations to a third party. The heavy lifting still happens on the machine you control; your phone is just the window into it, wherever you happen to be standing.

If you run into a snag, the most common fixes are almost always one of three things: the PC is asleep, Tailscale dropped its connection on one side, or the IP/port typed into LMSA doesn't match what your server is actually using. Check those three first, and you'll usually be back in business within a minute.