Start server (llama.cpp)

Common usage below — exact flags follow your llama.cpp version README.

Launch example

bash

./llama-server -m /path/to/model.gguf --host 0.0.0.0 --port 8080

Key points:

--host 0.0.0.0: allow LAN (phone cannot reach 127.0.0.1-only)
--port: custom port; allow it in the firewall
Some versions need an extra flag for OpenAI-compatible routes (recent servers usually ship /v1/chat/completions)

text

http://127.0.0.1:8080/v1/models
http://YOUR_PC_IP:8080/v1/models

Field	Example
Base URL	`http://YOUR_PC_IP:8080/v1`
API Key	Empty or placeholder
Model name	Same id as `/v1/models` (some builds use the filename)