Google Gemini API with PicoClaw
Gemini is supported as a first-class provider in PicoClaw. Use it for assistants that need Google’s multimodal models while keeping orchestration lightweight on Raspberry Pi, VPS, or homelab hosts.
1. Get an API key
Create a key in Google AI Studio. Treat it like any production secret: environment variables, locked-down config files, or your host’s secret store—not git.
2. Configure PicoClaw
Add the gemini entry under providers and set your default model in agents.defaults.model (or per-agent overrides). Field names and examples are on Configuration; the provider table is on Providers.
3. When Gemini fits
- You already standardise on Google Cloud or AI Studio for keys and quotas.
- You want to compare outputs against OpenAI or Groq for the same PicoClaw workflow.
- You run automation on small hardware and only pay for inference in the cloud.