Google DeepMind's Gemma-4-31B-it-assistant and NVIDIA's Gemma-4-31B-IT-NVFP4 models are now accessible through Amazon SageMaker JumpStart. These models provide the powerful Gemma 4 31B dense architecture for enterprise use cases. The assistant-tuned Gemma-4-31B-it-assistant excels in multimodal reasoning, coding, and agentic workflows. It accepts text and image inputs, supports a vast context window, and handles over 140 languages. This variant ranks highly on open model leaderboards, outperforming larger models. It utilizes a hybrid attention mechanism and native function calling for autonomous agents. NVIDIA's Gemma-4-31B-IT-NVFP4 offers the same capabilities but with a significantly reduced memory footprint. Quantized to 4-bit FP4 precision using NVIDIA's ModelOpt, it requires less memory and offers faster inference. This optimized version maintains high quality and is suitable for cost-effective production deployments. SageMaker JumpStart allows users to deploy these models easily with minimal clicks. Customers can access these models via the SageMaker console or the SageMaker Python SDK.