Skip to content

Set up models and keys

Choose which models Chat can use, store their API keys safely, and check that they answer. Keys are kept in your operating system’s keychain, never in the vault.

New chats use Rhizome’s own engine (ADR-0187). Set up the models below in Settings → AI Agents.

Settings → AI Agents has three tabs. Local model is for models on your computer. API model is for hosted models and holds Taproot.

KindExamplesNeeds
HostedOpenAI, Anthropic, Gemini, OpenRouter, Groq, MistralAPI key, model ID
LocalOllama, LM StudioBase URL, model ID, usually no key
Your ownAny OpenAI-compatible serverBase URL, model ID, optional key and headers

Anthropic models are not offered by Rhizome’s own engine yet.

Under API key storage, pick one per provider:

  • Save locally in Rhizome (recommended). Paste the key. Rhizome stores it in the system keychain: macOS Keychain, Windows Credential Manager, or Secret Service on Linux. Keys that older versions saved in a settings file move to the keychain the first time they’re read.
  • Read from env var. Give the variable’s name. Rhizome reads the key from it.
  • No key. For local servers that don’t need one.

Don’t edit a key file by hand.

Click Test model on the provider. It sends one short request with no tools. If it passes, Settings says “Connection works. The model replied successfully.”

Taproot, the free automatic router on the API model tab, lets Rhizome pick free hosted models for you and move on when one runs out. Each free provider needs its own key. See Free models and routing.

Choose a model, or Taproot, in the picker. The choice is saved with your settings.

See Model provider connection.