# AWS adds full and quantized Gemma 4 variants to JumpStart

_Published Monday, September 14, 2026 at 5:16 PM EDT · AI · Latest · Tier 2 — Notable_

AWS says Google DeepMind's Gemma-4-31B-it-assistant and NVIDIA's Gemma-4-31B-IT-NVFP4 are now available on SageMaker JumpStart. The assistant-tuned model accepts text and image inputs, including video as frame sequences, and produces text, according to AWS. The company says the quantized NVIDIA version uses 4-bit FP4 precision, reducing memory use to about 18.5 GB versus the base model. AWS says customers can deploy both variants through JumpStart's model catalog or Python SDK.

## Sources

- [AWS](https://aws.amazon.com/about-aws/whats-new/2026/01/gemma-4-31b-it-assistant-gemma-4-31b-it-nvfp4-jumpstart/)

---
Canonical: https://techandbusiness.org/newswire/qPQQ1QFo6pNe0_t8bHF3vv
Published: 2026-09-14T21:16:24.692Z
Story chronology: 2026-09-14T17:32:00.000Z
Retrieved: 2026-10-30T00:34:55.898Z
Publisher: Tech & Business (techandbusiness.org)