🔗Providers
Boxxy connects to LLM providers via rig. Boxxy supports all models from each provider — automatically configuring thinking levels, context windows, and token pricing.
Configure your API keys or custom endpoints in =Preferences= → =APIs=, then assign models to each role (AI Chat, Claw, and Memories & Dreams) in =Preferences= → =Models Selection= or via the quick links in the =Claw Sidebar=.
🔗AI Model Providers
| Provider | Supported Models | Notes |
|---|---|---|
| Google Gemini | All Gemini models | Thinking level controls · Built-in Google Search |
| Anthropic Claude | All Claude models | Thinking controls · Built-in Web Search |
| OpenAI | All OpenAI models | Reasoning effort controls · Built-in Web Search |
| DeepSeek | All DeepSeek models | Chat & reasoning models · 1M context |
| Ollama | Any local model | Auto-discovers installed models on your machine |
| OpenRouter | Any OpenRouter model | Enter any model slug (e.g. google/gemma-4-31b-it) |
| Custom Endpoints | Any OpenAI-compatible model | Presets for OpenCode Go, OpenCode Zen, LiteLLM, and vLLM / llama.cpp with /models discovery |
🔗Web Search Providers
Characters can browse the web in real time to look up documentation, recent releases, or error codes. Enable web search in =Preferences= → =APIs= under Search Engines, and toggle it anytime per pane using the Globe Icon in the message bar (Ctrl + /).
| Search Provider | Supported Models | Notes |
|---|---|---|
| Native Provider Search | Gemini · OpenAI · Anthropic | Uses the provider's built-in server-side search — no extra API key required |
| Tavily | All providers | LLM-optimized web search · Automatic fallback for local and custom models |
More web search providers are coming soon!
🔗Missing a provider?
New model and search providers are added regularly. If there is a provider you would like to see supported, open an issue on GitHub!