Custom endpoints

Ollama, LM Studio & self-hosted AI on Android

Connect EchoFlow to OpenAI-compatible endpoints on your network. Chat with home-server models from your phone.

EchoFlow is an OpenAI-compatible Android client for self-hosted AI Android workflows. Use it as an Ollama Android client or LM Studio Android client front-end when your server is running and reachable.

EchoFlow custom endpoint configuration on Android

Typical setups

  • Ollama on a home PCconnect Android to Ollama over Wi‑Fi
  • LM Studio — remote LM Studio client while the desktop app serves models
  • Custom vLLM / LiteLLM / corporate gateways — any OpenAI-compatible base URL
  • Local network AI chat — keep inference on hardware you control

Networking and security

Your phone must reach the server IP and port. Firewalls, VPNs, and guest Wi‑Fi isolation can block connections—test on the same subnet first. Treat a home server AI client Android setup like any LAN service: trusted network, no public exposure without hardening.

Not the same as on-device models

Local models run on the phone itself. Custom endpoints run on another machine and stream results to EchoFlow—useful when your PC has a GPU but your phone does not.

Common questions

Can I use Ollama from my Android phone?
Yes. EchoFlow can connect to an Ollama server on your network using an OpenAI-compatible endpoint. Your phone and server must be on the same LAN or reachable securely—see our Ollama guide.
Does EchoFlow work with LM Studio?
Yes. Point EchoFlow at LM Studio's local OpenAI-compatible API when LM Studio is running on your PC and reachable from your phone.
Is exposing a local server safe?
Only expose inference endpoints on trusted networks. Do not port-forward Ollama or LM Studio to the public internet without proper authentication and TLS. We explain setup risks in our guides.

Try EchoFlow on Android

Free, open-source, and built for users who want multiple models, their own API keys, and chats that stay on their phone.