Burn Book runs a light-weight large-language model (LLM) directly on your iPhone thanks to MLX-Swift, so every roast is generated privately, instantly, and without racking up API fees.
App Store Link - https://apps.apple.com/us/app/burn-book-one-tap-roast/id6746701571
- 🎯 Target anything - Auto-detect, or tell the model to roast a Person or an Object.
- 🗡️ Roast Intensity - From friendly banter to “Homicidal” or an overkill heat—slide to adjust the cruelty level.
- 🏛️ Political toggle - Keep politics out, or lean all the way in. Your call.
- 🫧 Profanity filter - Ship-ready by default—switch it off if your audience can handle it.
- 📴 Offline, real-time - The 4-bit quantized model lives on device, no network, no wait.
- 📤 Share sheet - Drop burns straight into Messages, TikTok captions, or wherever shade is needed.
-
SwiftUI - Pure declarative goodness.
-
Core Haptics - Tactile feedback thrills every tap, slider move, and punchline.
-
AVFoundation - AVSpeechSynthesizer API reads each roast aloud. It lets you choose from multiple accents or even your Personal Voice.
-
LLM Runtime - MLX‑Swift wrapping the Metal‑accelerated MLX tensor library.
-
Model - Llama 3‑Instruct‑1B (~700 MB on disk), quantized to 4‑bit.
- Clone this repository:
git clone https://github.com/thisisayushs/BurnBook.git) - Open the project in Xcode.
- Run the app on a simulator or connected device.
This project is licensed under the MIT License. See the LICENSE file for details.