Set Up a Provider

A provider is the model service that answers your agent's messages.

Choose a route

Route	Use it when
`Cloud API`	you have an API key from Anthropic, OpenAI, OpenRouter, Groq, or another cloud provider
`Ollama`	you want Eliza fully local with no API key
`Eliza Cloud`	you want a managed backend with no local setup

Common cloud providers:

OpenAI
Anthropic
OpenRouter
Gemini
Groq
xAI

For detailed setup instructions, environment variables, and all 18 supported inference backends, see the Model Providers reference.

Connect it in Eliza

Open Settings.
Open the provider or AI model section.
Pick the provider.
Paste the API key, sign in, or enter the local server URL.
Run the connection test.
Pick a model if needed.

Ollama

Make sure Ollama is running before you test it. Verify with:

bash

curl http://127.0.0.1:11434/api/tags

If the model is not available yet, create it from the Eliza-1 Modelfile:

bash

ollama create eliza-1-9b -f packages/training/cloud/ollama/Modelfile.eliza-1-9b-q4_k_m

To see which models you already have: ollama list.

Remote backends

If you are connected to a remote Eliza backend, that backend may already own provider routing. In that case, changing a local key may do nothing. See Connect Remotely.

Switching later

You can change providers later if you need a different balance of:

cost
speed
privacy
model quality

Eliza keeps the same chats and agent state while the route changes.

Common problems

Messages hang

Check:

the API key
provider status
Ollama process status
the selected model name

Voice works but chat does not

Chat and voice can use different providers. Fix the chat route separately.

A remote or cloud backend ignores your local key

Confirm which machine owns provider routing.

All 18 providers

See Model Providers for the full list of supported providers, env variables, fallback chains, and advanced configuration.

Change Your Character