google_gemma-3-4b-it-Q4_K_M.gguf

Gemma 3 4B Instruct — Q4_K_M GGUF

Re-hosted for use with VaultCuisine, a privacy-first, offline recipe scanning app for Android.

This is a mirror of the Q4_K_M quantized GGUF build of Google's Gemma 3 4B instruction-tuned model, used as the offline/on-device inference tier in VaultCuisine. It structures OCR-extracted recipe text (ingredients, quantities, steps) into structured JSON entirely on-device, with no data sent to any server.

Source

• Original model: google/gemma-3-4b-it by Google DeepMind
• GGUF quantization: bartowski/google_gemma-3-4b-it-GGUF
• This file is an unmodified re-host of the Q4_K_M quant for direct download by the VaultCuisine app.

Quantization: Q4_K_M (~2.5 GB) — chosen as the best balance of accuracy and on-device memory footprint for mid-range and flagship Android hardware.

Intended use: Local, offline text structuring within VaultCuisine's optional "Offline Mode." Not intended as a general-purpose chat or instruction-following endpoint.

License

Gemma is provided under and subject to the Gemma Terms of Use found at ai.google.dev/gemma/terms. By downloading or using this model, you agree to be bound by that agreement, including the Gemma Prohibited Use Policy. This repository is a redistribution of Model Derivatives as permitted under Section 3 of the Gemma Terms of Use; no modifications have been made to the model weights.