im.ai is a privacy-first AI chat app for iOS and macOS that runs large language models locally on-device or lets you bring your own API keys from providers like Google Gemini, Groq, OpenRouter, Cerebras, and GitHub Models — no accounts, no server, no prompt logging, and completely free.
What is im.ai?
im.ai is a native app for iOS 18.6+ and macOS 15.7+ that allows you to chat with large language models either entirely offline using local inference via llama.cpp or by connecting to cloud providers with your own API keys. It accepts text prompts and returns AI-generated responses. The app is built by a solo developer and runs on both Apple Silicon and Intel Macs.
Key Features
- Local on-device inference — Runs Llama, Gemma, Qwen, and Falcon models directly on your iPhone or Mac using llama.cpp; no data ever leaves your device.
- Bring your own API keys — Supports Google Gemini, Groq, OpenRouter, Cerebras, and GitHub Models; keys are stored securely in Apple Keychain.
- Unified interface — Switch between local and cloud models with a single tap; consistent chat UI across all providers.
- No accounts or servers — No sign-up, no backend, no prompt logging. Your conversations stay private by design.
- Free with no in-app purchases — Available on the App Store with a free TestFlight build for early access.
- Solo-developed — Built by a single developer, with optimizations for both Apple Silicon and Intel processors.
Who is it for?
- Privacy-conscious individuals who want to use AI without sending data to third parties; run models entirely offline for sensitive questions.
- Developers and power users who already have API keys from supported providers and want a single app to access all their models without maintaining multiple clients.
- iOS and macOS users on iOS 18.6+ or macOS 15.7+ looking for a free, private AI chat tool that works locally or with their own cloud keys.
What can you do with im.ai?
- Offline question answering — Use local models like Llama to get answers without an internet connection, ideal for travel or handling confidential data.
- Compare providers mid-conversation — Run the same prompt across local models and multiple cloud providers (Gemini, Groq, etc.) to see which response fits best.
- Test and benchmark models — Using your own API keys, evaluate performance and quality across different models from a single interface.
How does im.ai work?
- Download the free app from the App Store. 2. Choose to use local models (downloaded on-device) or add your own API keys from supported providers via Apple Keychain. 3. Start chatting in the unified interface; switch between local and cloud models at any time. 4. No data is logged — your prompts go directly to the chosen provider or stay local.
Pricing
im.ai is completely free with no in-app purchases. There is also a TestFlight build available for early access.
FAQ
Is im.ai free?
Yes, im.ai is free with no in-app purchases or subscriptions. A TestFlight version is available for early adopters.
What models can I run locally?
The app supports Llama, Gemma, Qwen, and Falcon models via llama.cpp. These are downloaded and run entirely on your device.
Can I use my own API keys?
Yes. You can bring your own keys for Google Gemini, Groq, OpenRouter, Cerebras, and GitHub Models. Keys are stored in the Apple Keychain.
Is my data private?
Absolutely. There are no accounts, no servers, and no prompt logging. Local models keep everything on-device; cloud requests go directly to your chosen provider without an intermediary.
What platforms are supported?
im.ai requires iOS 18.6+ and macOS 15.7+. It runs on both Apple Silicon and Intel Macs.
