Model providers

Choose a provider

Usage and billing go through your provider account. A subscription sign-in and an API key from the same company are separate connections.

Connect a provider

In the desktop app, open Settings → Server → Provider connection, choose a provider, and connect.

Or, on the server host, run guided setup:

openteam setup

If you've already signed in to Codex or Claude Code on the host, setup can reuse that sign-in.

Choose a model

In the desktop app, open Settings → Server → Model and reasoning, pick a model and reasoning effort, and choose Apply. Or run openteam model on the host for an interactive picker.

All bots use the same model. Changes apply to new messages; a task that's already running finishes with the settings it started with.

New installations have no saved model. After connecting a provider, the model picker defaults to the first available model in its catalog. Save or apply the selection to use it for new tasks. Existing saved model choices are preserved, even when the provider changes its catalog order.

Higher reasoning effort can give better results on hard tasks, but it's slower and uses more of your quota. New installations start at medium. Some models don't support reasoning effort, and the setting is unavailable for them.

To script the change:

openteam model list
openteam model use <provider> <model> --thinking medium

Use OpenRouter

Choose OpenRouter and paste your API key. The model list shows the models your OpenRouter account allows that support tool use. OpenRouter model IDs include the model's author, such as author/model-name.

Use your own model server

You can connect any server that speaks one of these APIs: OpenAI Chat Completions, OpenAI Responses, Anthropic Messages, or Google Generative AI. Bots rely on tool calling, so choose a model that supports it.

The server must be reachable from the bots' computer, which runs in Docker. Inside Docker, localhost means the container, not your host. Use host.docker.internal to reach a server running on the host, such as Ollama. On Linux, the server must listen on more than 127.0.0.1; for Ollama, set OLLAMA_HOST=0.0.0.0.

openteam provider add local --name "Local models" \
  --base-url http://host.docker.internal:11434/v1 \
  --api openai-completions --no-auth
openteam model list local
openteam model

If your endpoint needs a key, leave out --no-auth and the CLI will ask for it. The --api option accepts openai-completions, openai-responses, anthropic-messages, or google-generative-ai.

Check the connection

Run openteam doctor to send a test request to your model. A model can appear in the list and still fail if your account has no quota left.

If your sign-in expires, choose Disconnect and connect again in Settings → Server, or run openteam provider login.

Web search and voice transcription use their own credentials. See web search and voice notes.