google/gemma-3-4b-it-qat-q4_0-unquantized

The google/gemma-3-4b-it-qat-q4_0-unquantized model is a 4.3 billion parameter instruction-tuned variant from Google's Gemma 3 family of lightweight, open multimodal models. Built with technology from Gemini models, it handles text and image inputs, generating text outputs, and features a 32K token context window. This specific version is optimized using Quantization Aware Training (QAT) to maintain quality while reducing memory footprint, making it suitable for resource-constrained environments and a variety of text generation and image understanding tasks.