Downloads

Apps and offline resources. Sizes are downloads.
Resource	Size	Download
Android app
APK not yet available.
Terminal app
Windows x64 Details v0.2.0 Filename:`prep-cli-v0.2.0-windows-x64.zip` SHA-256:`3adc2958bc87ff9b9535b00ad2ccac8b895dcf1201951bb5e9770ca63d39ec44`	33.3 MB	Download
Linux ARM64 Details v0.2.0 Filename:`prep-cli-v0.2.0-linux-arm64.zip` SHA-256:`1881b444f151f804f3c6134e6dc1712708d9a8fd91014e55b1ecade926f4f898`	41.6 MB	Download
Linux x86_64 Details v0.2.0 Filename:`prep-cli-v0.2.0-linux-x64.zip` SHA-256:`923d33a06ecd0152f4fc1ae907c1497bbf2a6ac21e2f1f9148c3c6a9205b5139`	41.7 MB	Download
Offline maps
New York City Local streets, paths, buildings, and landmarks in New York City and nearby areas, where mapped. Details Map data date 2026-10-20 Protomaps · OpenStreetMap contributors · Natural Earth OpenStreetMap attribution and license In prep.zip, open Places → Offline maps and import this map package. After installation, the map works without a network connection. Map data dated October 20, 2026. Does not include route calculation, satellite imagery, or live conditions. Filename:`new-york-city-sample.prepmap` SHA-256:`633787e6fa08e8a345fd49eef61a0aaf39cedfcee3802f5f898b402f9cb07f6e`	95.6 MB	Download map
Food databases
USDA food reference Nutrition and reference portions for 13,500+ foods and ingredients. Details Installed size 23.4 MB Version 1 USDA FoodData Central · CC0 Foundation: April 2026 · SR Legacy: April 2018 · FNDDS: October 2024 Filename:`usda-food-v1-e8aac4c073798a61.sqlite.gz` SHA-256:`ee9e662386ac580c66ef4e865f08490c01e896f3a3ada711041a49734129b701`	8.0 MB	Download database
USDA branded food reference Nutrition, package details and barcode lookup for 435,000+ branded products. Details Installed size 356.3 MB Version 1 USDA FoodData Central · CC0 Branded Foods: April 2026 Filename:`usda-branded-v1-fe5132506bca2172.sqlite.gz` SHA-256:`213ce6f91079c32339f53b9c6554e8ef67f9f5bb4c9e909b24d397fca4a83711`	153.7 MB	Download database
Assistant models Install and manage models in prep.zip Assistant settings. Models download from Hugging Face and run in a supported browser on your device.
Light · Smol 135M Lowest GPU memory use for simple responses. Details Hugging Face Models also need working memory and free storage during installation.	277.3 MB	Install in app
Balanced · Smol 360M General-purpose responses with moderate GPU memory use. Details Hugging Face Models also need working memory and free storage during installation.	212.3 MB	Install in app
Advanced · Qwen3.5 0.8B High-quality responses with high GPU memory use. Details Hugging Face Models also need working memory and free storage during installation.	444.1 MB	Install in app
Max · Qwen3.5 2B Most capable responses with highest GPU memory use. Details Hugging Face Models also need working memory and free storage during installation.	1.08 GB	Install in app