Settings
Providers
Connect built-in providers, local providers, OAuth-backed providers, and custom compatible endpoints.
Built-in providers
The Providers panel lists the built-in provider rows that the native app knows how to configure:
- OpenAI.
- ChatGPT Plus/Pro.
- OpenRouter.
- Anthropic.
- Gemini.
- GitHub Copilot.
- Ollama.
- LM Studio.
- OpenCode Go.
- OpenCode Zen.
Some use API keys, some use OAuth/device auth, and local providers can run without an API key.
Custom providers
Add Custom Provider supports:
- OpenAI-compatible endpoints.
- Anthropic-compatible endpoints.
You provide a display name, base URL, and optional API key. Blank API keys are allowed for local OpenAI-compatible servers.
Model discovery
Built-in providers fetch remote model lists where supported. The fetched list is also used to infer capabilities such as vision support and context window when possible.
If a model is missing, add it manually as a custom model under that provider.
Active selection
The chat model selector stores the active provider and model pair. Task dispatch can also set the active selection before starting a native dispatched session.
Where keys live
Provider configuration is stored in the local AgentsOS database. AgentsOS does not proxy requests through an AgentsOS cloud service.