Connect Ollama or LM Studio to a Phone
Short version
Local-server mode lets your phone send prompts to a model endpoint you run yourself, such as Ollama, LM Studio, or another OpenAI-compatible server. This gives you a mobile interface while heavier models run on your own computer.
Why use a local server
A desktop or laptop can run larger models than a phone. Local-server mode gives you a calm mobile interface while the heavier model runs on hardware you control.
This can be useful for longer planning sessions, code help, startup work, study material, or larger reading tasks.
Basic setup
Make sure your phone and computer are on the same trusted network. Start your server software and confirm it accepts connections from other devices on the network.
In Phos, add the server URL, test the connection, and select the model exposed by that server.
Keep the boundary clear
Do not expose local model servers directly to the public internet unless you know how to secure them.
Local-server mode is private only to the extent that your network, server, and model software are private. Phos treats this as a user-run endpoint, not company-hosted inference.
When local server beats phone-only local
Use a local server when:
- the model is too large for your phone,
- you want a longer context window,
- your phone overheats with local inference,
- you want desktop hardware but mobile UX,
- you already run Ollama or LM Studio.
Use phone-only local mode when the thought is sensitive and the local model is good enough for the task.
FAQ
Can I use Ollama from an Android phone?
Yes. If Ollama is running on a reachable computer or server, an Android app can connect to that local endpoint over a trusted network.
Is local-server mode the same as on-device local mode?
No. On-device local mode runs on the phone. Local-server mode sends the prompt to the server URL you entered, usually on your own Wi-Fi network.
Start with a private setup
Phos can run locally, connect to your own server, or use your own provider key when you choose.