Your AI, your rules.
Local models when you want privacy. Cloud providers when you need power. Switch freely, no lock-in, no sign-in.
Llama, Gemma, Qwen and more run directly on your device via llama.cpp. Nothing leaves your iPhone or Mac.
Your own API keys, direct connections to Google, Groq, OpenRouter, Cerebras, and GitHub Models. No middlemen.
Native markdown with code blocks, full session history, and instant model switching mid-conversation.
Hardware-accelerated inference, low memory footprint. Snappy whether you're running on-device or hitting a cloud API.
Text size, reasoning effort, markdown rendering, spell check, every detail configurable. API keys stay in the system Keychain.
Enable reasoning mode on compatible models for step-by-step thinking on harder problems. Works locally and with cloud providers that support it.
Crafted for iOS and macOS with full light and dark theme support.
Full markdown with code blocks. Ask anything. Responses are clear, readable, and beautifully formatted.
Run the same prompt across multiple models and see answers side by side. Pick the best for the job.
Switch between Local, Google, Groq, OpenRouter, Cerebras, and GitHub. Any prompt, any time, no lock-in.
Access Gemma, Llama, Qwen, and free cloud tiers. Many models need no API key at all. Start exploring immediately.
All conversations saved, searchable, and organized by date. Context carries forward so you can pick up where you left off.
Markdown rendering, reasoning effort, text size, spell check, every detail is yours to configure.
Models run entirely on your device via llama.cpp. No network connection is made. Nothing leaves your iPhone or Mac.
Bring your own API key. im.ai connects directly from your device to Google, Groq, OpenRouter, Cerebras, or GitHub Models. No relay, no proxy, no im.ai servers in the path.