Full threadtpae·I've been building with local AI, on Apple Silicon. It's only 8mb, but runs 30% faster than Ollama.https://github.com/dinoki-ai/osaurusView on HN