The Short Version
Everything runs on your iPhone. Quanta collects nothing, sends nothing, and has no servers.
Your conversations: never transmitted
All AI processing happens on-device using Apple's built-in model and/or open models you download (run via MLX). Your prompts, attached images, and replies are processed locally and never sent anywhere. After a model is downloaded, Quanta works fully offline.
Servers: none
Quanta has no backend. There is no account, no user database, and no way for us to see your chats — because we never receive them.
Voice & images
Voice input is transcribed on-device using Apple's Speech framework (on-device recognition). Photos you attach are analyzed locally by an on-device vision model and are never uploaded. Microphone audio and images are processed in memory for your request and are not transmitted.
Analytics & tracking: none
Quanta includes no analytics, crash trackers, advertising SDKs, or tracking pixels. We have no idea how you use the app — by design.
Local storage
Conversations and preferences are stored only on your iPhone. Downloaded models are stored locally. You can delete any conversation or model at any time from within the app, or delete everything by removing the app.
Model downloads
When you choose to download an open model, the app fetches it directly from the model host (Hugging Face). That request contains only what's needed to download the files (no chat content). This is the only network activity Quanta performs, and only when you initiate a download.
Purchases
The optional one-time unlock is processed by Apple via the App Store using StoreKit. We never see your payment details. Apple's privacy policy governs that transaction.
Children
Quanta is a general-purpose tool and does not knowingly collect any data from anyone, including children.
Contact
Questions? Email info@visionion.dev.