Radiant - Local AI Chat

Radiant runs open-weight models directly on your iPhone.

Download a model once and it works anywhere — on a plane, underground, with no signal at all. Nothing you send it leaves the device, because there is nowhere for it to go: no account, no sign-in, and no server of ours in between.

CHOOSE FROM 53 MODELS

Models from Google, Meta, Mistral, Microsoft, IBM, Alibaba, NVIDIA, DeepSeek, Liquid AI, Hugging Face and more — grouped by who made them, from 0.2 GB up.

Every model is labeled for YOUR iPhone before you download it: Runs well, Runs tight, or Won't run. That verdict is measured against the memory iOS actually grants an app on your specific device, not a guess from the spec sheet — so you know before you spend the download.

BUILT FOR THE PHONE

· Conversations are kept and named, so you can pick one up later
· Switch models mid-conversation without losing what you were saying
· Twelve color themes, four appearance modes, and your own text size
· Full Dynamic Type and VoiceOver support

BRING YOUR OWN KEY, OPTIONALLY

If you want a model too large for any phone, add your own API key for Anthropic, OpenAI, OpenRouter, xAI, Nous, DeepSeek, Kimi, GLM, Groq or Mistral. Keys are held in the iOS Keychain. The line under every chat title tells you which model is answering and where it runs, so you always know whether you are on-device or online.

Radiant is a Templeton Technologies product.