Ollama is the $0 default
Ollama runs models locally, so there is no provider bill and no key to manage. It is the obvious choice for private or routine work.
Wire it up
- Install Ollama from ollama.com.
- Pull a model:
ollama pull llama3.1. - In the console, set the provider to Ollama with the local endpoint.
- Pick the model and save — no key field to fill.
When Ollama is enough
For summaries, file work, simple automation, and anything sensitive, a local model is plenty. It is slower and less sharp than a frontier model, but it is free and private.
Mix and match
You can run Ollama for private tasks and keep a provider key in the vault for the hard ones. The auto-provider can route each task to the cheapest fitting key.
Check it is working
$ remoteagent chat "what model are you running?"