About Mmodel

Mmodel Run GGUF LLMs on your Android phone — find models that fit, download them, and chat fully offline.** Mmodel = **m**obile **model**. It reads your phone's real RAM, storage and CPU, checks the actual file sizes published on Hugging Face, and tells you which quantizations will **work**, be **slow**, or **won't run** — before you spend gigabytes downloading. ⬇ Download the latest APK**  ·  Features  ·  Install  ·  How fit works  ·  Build  ·  FAQ Why Mmodel? Most "AI on Android" apps either stream from a server or dump a model file on you and hope for the best. Mmodel answers the question that actually matters first: **Will this model run on *my* phone?** **Real numbers, not guesses.** Fit is computed from the actual GGUF file sizes Hugging Face publishes, including every shard of a quantization — not from parameter counts. **Private by design.** No account, no analytics, no tracking, no model server. Prompts and replies never leave the device. The app requests only the INTERNET permission. **One flow from discovery to chat.** Search Hugging Face, pick a quantization, download it with pause/resume, then talk to it in the same app. **Open source and reproducible.** Kotlin + Jetpack Compose. Every push to main is tested and built in GitHub Actions, and a signed APK is published automatically. Features **Live device profile** —…

Release archive