How to start the Ollama server

The Ollama server provides the local HTTP API that CLI commands, SDKs, browser apps, and integrations use. Starting it explicitly is useful on headless hosts, in containers, or when the desktop app is not managing the runtime.

Foreground ollama serve is convenient for testing environment variables. Production Linux installs normally use the ollama systemd service so the server starts on boot.

Check the API endpoint after starting the server. A process that starts but does not answer on the expected host and port will still fail client requests.

Steps to start the Ollama server:

Start Ollama in the foreground for a temporary test.
```
$ ollama serve
Listening on 127.0.0.1:11434
```
Start the Linux service for a service install.
```
$ sudo systemctl start ollama
```
Related: How to manage the Ollama service

Check that the API endpoint answers.

$ curl -s http://localhost:11434/api/version
{"version":"0.31.1"}

Set server environment variables before starting when needed.
```
$ OLLAMA_HOST=0.0.0.0:11434 ollama serve
```
Related: How to set the Ollama host address
Check logs if the server exits immediately.
```
$ journalctl -e -u ollama
##### snipped #####
```
Related: How to view Ollama logs