im.ai is a small, privacy-first AI app for iOS and macOS. Two ways to run a model, and both keep the middleman out.
Local models, on-device. Run Llama, Gemma, Qwen, Falcon, and others directly on your iPhone or Mac via llama.cpp. No internet required, and nothing leaves your device.
Bring your own keys. Connect straight to Google Gemini, Groq, OpenRouter, Cerebras, and GitHub Models with your own API keys — no relay servers in between. Keys are stored in the system Keychain.
Key Features
- On-device inference — Llama, Gemma, Qwen, Falcon, and more via llama.cpp.
- Direct provider connections — Gemini, Groq, OpenRouter, Cerebras, GitHub Models, with your own keys.
- Native markdown rendering — Including code blocks.
- Full session history — Conversations persist across launches.
- Side-by-side model comparison — Run the same prompt through two models at once.
- Instant model switching — Change models mid-chat without losing context.
- Reasoning mode — On models that support it.
Who It’s For
People who want a chat client that either runs entirely offline or talks straight to a provider — no intermediary service seeing the conversation, and no subscription on top of the API costs they are already paying.
Requirements
A universal app for iOS 18.6+ and macOS 15.7+ (Apple Silicon and Intel). Free to download from the App Store.