How to launch an Ollama integration

The ollama launch command configures supported external developer tools so they can use Ollama models. It is useful when a coding assistant or editor needs provider settings without manual environment-variable setup.

The current CLI lists integrations such as OpenCode, Claude Code, Codex, VS Code, and Droid. An interactive launch can configure and start a tool, while --config prepares settings without opening the integration.

Run the command from a terminal that has access to the target tool. If a tool is not installed, use the generated configuration as a checklist rather than treating launch failure as an Ollama server problem.

Steps to launch an Ollama integration:

Confirm Ollama is running before configuring an external tool.

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

Start the interactive integration launcher.

$ ollama launch
? Select an integration  [Use arrows to move]
  Claude Code
  Codex
  VS Code
  Droid

Launch a specific supported integration when the target is known.
```
$ ollama launch codex --model gpt-oss:20b
```
Generate configuration without launching the external app.
```
$ ollama launch droid --config
```
Run a simple model request after the integration starts.
```
$ ollama run gpt-oss:20b "Return only OK."
OK
```