Connections
Start with Settings → Server → Test Connection. Read the result, then check one cause at a time.
Test Connection fails
Section titled “Test Connection fails”- Confirm LM Studio’s API server is running and the host is awake. A loaded model or running daemon alone does not prove the API server has started.
- Compare Server URL with the address and port shown by LM Studio. Use the base URL, without an API path.
- Enable Serve on Local Network and confirm your iPhone or iPad can reach the host computer. Guest Wi-Fi can isolate clients even if the network name looks correct.
- If authentication is enabled, check Auth Key against a valid token. Do not prefix the saved token with
Bearer. - Test again after each change.
The server cannot be reached
Section titled “The server cannot be reached”Use the host computer’s network address from your iPhone or iPad. localhost and 127.0.0.1 refer to the mobile device running LLMster and cannot reach your computer.
Allow Local Network access for LLMster in system Settings if it was denied. On the host, allow incoming connections on the server’s configured port. Check whether a VPN or guest network prevents local access.
Authentication is rejected
Section titled “Authentication is rejected”Confirm Require Authentication and the token’s permissions in LM Studio. A token can be revoked or restricted even when its text is saved correctly in the app. Create a replacement on the host if necessary, enter it in Auth Key, and test again.
Do not include tokens in screenshots or support messages. Turning off authentication is not required to diagnose an incorrect address or denied network permission.
It worked before a reboot
Section titled “It worked before a reboot”The computer’s IP address may have changed, the server may not have restarted, or the host may have gone to sleep. Check those before changing the app. A router DHCP reservation can keep the host address stable.
Connected, but there is no reply
Section titled “Connected, but there is no reply”A successful connection test verifies access to model information. Check that a usable chat model is selected and loaded, then follow Models and responses.