01 / COGNITION
The Hybrid Brain
JARVIS keeps routine intelligence close to the metal. Qwen 2.5, Qwythos-9B, and Gemma run locally through Ollama on Apple Silicon—private, responsive, and available without a round trip to the cloud. When a mission demands deeper context or heavier reasoning, the same orchestration layer routes work to Google Cloud Vertex AI, Gemini, Claude, DeepSeek, OpenRouter, or your own provider.
Apple Silicon / Ollama
Cloud inference fabric
