Model Providers

Orbicus connects directly to model providers over encrypted connections without routing prompts through intermediate proxy servers. You can choose any supported cloud provider, an enterprise gateway, or a locally hosted model.

Supported Protocols

Orbicus implements native support for the primary AI API protocols:

Anthropic: Official Claude models (claude-3-7-sonnet, claude-3-5-haiku, claude-3-opus) and compatible endpoints.
OpenAI: Official models (gpt-4o, o3-mini, gpt-4o-mini) as well as third-party OpenAI-compatible providers.
Google Gemini: Gemini Developer API (gemini-2.5-pro, gemini-2.0-flash) and Vertex AI enterprise gateways.
Local & Offline Inference: Fully offline execution via Ollama, vLLM, or LM Studio.

Configuring Providers

Cloud API Keys

To configure a commercial cloud model:

Navigate to Settings $\rightarrow$ Providers.
Select your provider tab (Anthropic, OpenAI, or Google Gemini).
Enter your API key. Keys are encrypted and stored locally in your user profile (~/.orbicus/).
Select your preferred default model and set the maximum output token limit.

Local Models with Ollama

For environments requiring complete privacy or offline operation, you can point Orbicus to a local model server:

Start your local Ollama daemon:
```
ollama run qwen2.5-coder:14b
```
In Orbicus Providers settings, select Ollama / Local.
Set the endpoint address to http://localhost:11434.
Pick your installed model from the detected model list.

Custom Endpoints & Proxies

If you use an OpenAI-compatible provider (such as Groq, Together AI, Mistral, DeepSeek, or OpenRouter) or an internal company API gateway:

Select Custom Provider in the settings panel.
Provide the base URL (for example, https://openrouter.ai/api/v1 or https://api.groq.com/openai/v1).
Enter your bearer token or API key.
Specify the exact model identifier to target.