Quantized Gemma 3 1B Instruct used by the Offline Answer Evaluation Engine (Dextora). Runs on-device via llama.cpp. ~769 MB (under 1 GB).

Used by the Phone-LLM service in dextora_micro_services. The backend downloads this file at startup. See that repo for the API and Flutter integration guide.

File: gemma-3-1b-it-Q4_K_M.gguf