Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.lmsa.app/llms.txt

Use this file to discover all available pages before exploring further.

Most LMSA issues fall into one of a few categories: a local server that isn’t reachable, a misconfigured API key, or a model that hasn’t been loaded yet. Work through the relevant section below to resolve your issue. If you’re still stuck, email support@lmsa.app.
The most common cause is a network or configuration mismatch between your phone and your PC. Work through these steps in order.
1

Confirm the server is running

LMSA connects to a server that you start on your PC — it does not start the server for you.
  • LM Studio: Open LM Studio, load a model, and click Start Server in the Local Server tab. The status should show a green indicator and a port number (default: 1234).
  • Ollama: Ollama runs as a background service. Confirm it is active by opening a terminal and running ollama list. If it is not running, start it with ollama serve.
2

Check that both devices are on the same Wi-Fi network

Your phone and PC must be connected to the same Wi-Fi network — the same router and the same SSID.
Guest networks and networks with different SSIDs (even on the same router) often block device-to-device traffic. If you are on a guest network, switch to the main network on both devices.
3

Find your PC's local IP address

You need your PC’s local IP address (not the public internet address) to configure LMSA.
  • Windows: Open Command Prompt and run ipconfig. Look for the IPv4 Address under your Wi-Fi adapter — it will look like 192.168.x.x.
  • Mac: Open Terminal and run ifconfig | grep "inet ", or go to System Settings → Network → Wi-Fi → Details and note the IP address.
Enter this IP address in LMSA under Settings → Connection → Configure.
4

Verify the correct port

Default ports are:
ProviderDefault port
LM Studio1234
Ollama11434
If you changed the port in your server settings, use the custom port you configured instead.
5

For Ollama: set OLLAMA_HOST to 0.0.0.0

By default, Ollama only listens on localhost (your PC itself) and will not accept connections from other devices on the network.To allow LMSA to connect, set the OLLAMA_HOST environment variable before starting Ollama:
  • Windows (PowerShell): 
    $env:OLLAMA_HOST = "0.0.0.0"
ollama serve
  • Mac / Linux (Terminal): 
    OLLAMA_HOST=0.0.0.0 ollama serve
On Windows, you can also set this as a permanent system environment variable in System Properties → Environment Variables so you don’t need to set it each time.
6

Check your firewall

Your PC’s firewall may be blocking the incoming connection from your phone.
  • Windows Defender Firewall: Open Windows Security → Firewall & network protection → Allow an app through firewall and ensure LM Studio or Ollama is allowed on private networks.
  • macOS Firewall: Go to System Settings → Privacy & Security → Firewall and check that incoming connections to LM Studio or Ollama are not being blocked.
You can also test connectivity by trying to open http://<your-pc-ip>:<port> in a browser on your phone. If you can reach it, the firewall is not the issue.
7

Check the LMSA status indicator

In LMSA, go to Settings → Connection. If the status reads Not configured, no IP address or port has been saved yet. Tap Configure to enter your server’s IP and port, then save.After saving, return to the main chat screen and try sending a message. If the connection succeeds, the model list will populate automatically.
1

Verify your API key

Go to openrouter.ai and sign in to your dashboard. Copy your API key directly from the dashboard and paste it into LMSA under Settings → Connection → OpenRouter.
Do not type the API key manually — copy and paste it to avoid typos. Even a single incorrect character will cause authentication to fail.
2

Check your OpenRouter credits or free quota

Some models on OpenRouter require credits. If your account has no credits and you have selected a paid model, requests will fail.To test whether your key is working, switch to a free model in the model selector (free models are labelled in the OpenRouter model list) and try sending a message.
3

Confirm the correct provider is selected

In LMSA Settings → Connection, make sure the OpenRouter option is selected, not Local Server or Custom Endpoint. Only one provider can be active at a time.
LMSA retrieves the model list from your connected provider at connection time. If the list is empty, the provider either has no model loaded or LMSA cannot reach it.
1

Load a model on your server first

LMSA cannot list models that have not been loaded or pulled yet.
  • LM Studio: Open LM Studio and load a model before starting the server. The server must be running with an active model for LMSA to see it.
  • Ollama: Run ollama pull <model-name> to download a model, then confirm Ollama is running with ollama list.
2

Verify the correct provider toggle

Go to Settings → Connection and confirm the active provider matches the server you are running. If LM Studio is running but OpenRouter is selected, LMSA will not find your local models.
3

Refresh the model list

After confirming the server has a model loaded, return to the LMSA chat screen and use the model selector to refresh. If models still don’t appear, check that the connection is working by reviewing the steps in the “Cannot connect” section above.
1

Increase Max Tokens

If responses are cut off mid-sentence, the Max Tokens setting may be too low. Go to Settings and increase Max Tokens, or set it to 0 to allow unlimited output length.
2

Confirm the model is still loaded

Local servers can unload models automatically after periods of inactivity. Check LM Studio or Ollama on your PC to confirm a model is still actively loaded. Reload it if necessary, then retry in LMSA.
3

Increase the Reasoning Timeout for thinking models

For models that perform extended reasoning (like DeepSeek-R1), LMSA applies a default timeout of 5 minutes. If the model is still working when the timeout expires, the response will be cut off.Go to Settings and increase the Reasoning Timeout value to give the model more time to complete its response.
Some models — particularly certain versions of Qwen — behave erratically when reasoning prompts are active. You may see the model repeat phrases like “wait, I should” dozens of times before producing a response.
This is a known behavior of specific model versions, not a bug in LMSA.
1

Disable reasoning for that model

Go to Settings and set Reasoning Level to Disabled. This prevents LMSA from sending reasoning-specific instructions to the model, which resolves the loop behavior for affected models.
2

Try a different model

If you need extended reasoning, switch to a model that is designed for it, such as DeepSeek-R1 or a model explicitly listed as a reasoning model by its provider.
The biometric unlock option in LMSA requires that your Android device has biometric authentication (fingerprint or face unlock) already configured at the system level.
1

Set up biometrics in Android settings

Go to your device’s Settings → Security → Biometrics (the exact path varies by manufacturer) and enroll at least one fingerprint or set up face recognition.
2

Enable biometric unlock in LMSA

Once biometrics are enrolled on your device, return to LMSA Settings → Security and enable Biometric Unlock. The option will be available as soon as a biometric is registered on your device.
If your device does not have a fingerprint sensor or face unlock hardware, biometric authentication will not be available regardless of settings.

​
Still need help?

If none of the steps above resolve your issue, reach out directly. Email support@lmsa.app with a description of the problem, your Android version, and which provider you are trying to connect to. The LMSA team responds to all support requests personally.