im.ai is a small chat app for iOS and macOS. It runs local models (Llama, Gemma, Qwen) on-device via llama.cpp, and can also connect to Google, Groq, OpenRouter, Cerebras, and GitHub Models using your own API keys. No accounts, no relay, no servers in the path. API keys live in the system Keychain. It's an early project — feedback welcome.
