OFICIAL AWS What's New

FLUX.2-small-decoder and gemma-4-12B-it models now available on Amazon SageMaker JumpStart

What happened
Based on AWS What's New · Aug 10, 2026

AWS has added Black Forest Labs' FLUX.2-small-decoder and Google's gemma-4-12B-it models to Amazon SageMaker JumpStart, expanding accessible AI tools for AWS customers.

FLUX.2-small-decoder and gemma-4-12B-it models now available on Amazon SageMaker JumpStart
AWS What's New — Amazon Web Services
Key points
·
Black Forest Labs' FLUX.2-small-decoder and Google's gemma-4-12B-it models are now available on Amazon SageMaker JumpStart, expanding the portfolio of foundation models available to AWS customers.
·
These two models bring specialized capabilities spanning efficient image generation decoding and unified multimodal understanding, enabling customers to deploy high-performance, scalable AI solutions on AWS infrastructure.
·
FLUX.2-small-decoder is optimized for faster image decoding with lower VRAM usage in FLUX.2 image generation pipelines.
·
It is a distilled VAE decoder that serves as a drop-in replacement for the standard FLUX.2 decoder, delivering approximately 1.4× faster decoding speed at 1.4× lower VRAM consumption with minimal to zero quality loss.
Key numbers
·
Google's gemma-4-12B-it model has also been added to SageMaker JumpStart, providing unified multimodal capabilities across text, image, and audio inputs.
·
It delivers performance comparable to Google's larger 26B MoE model while requiring less than half the memory, enabling deployment on systems with as little as 16GB of RAM.
·
2-small-decoder and Google's gemma-4-12B-it models to Amazon SageMaker JumpStart, expanding accessible AI tools for AWS customers.

Black Forest Labs' FLUX.2-small-decoder model is now available on Amazon SageMaker JumpStart, offering a faster and more efficient alternative to the standard FLUX.2 decoder. The model reduces VRAM consumption by approximately 1.4× while increasing decoding speed by the same factor, with minimal quality impact. These improvements are more pronounced at higher resolutions, making it suitable for large-scale image generation tasks in production environments.

Google's gemma-4-12B-it model has also been added to SageMaker JumpStart, providing unified multimodal capabilities across text, image, and audio inputs. The model supports function calling and agentic workflows, operating with an encoder-free architecture that consolidates modalities into a single decoder-only transformer. It delivers performance comparable to Google's larger 26B MoE model while requiring less than half the memory, enabling deployment on systems with as little as 16GB of RAM.

Both models are accessible through Amazon SageMaker JumpStart, allowing customers to deploy them with minimal setup. Users can locate the models in the SageMaker JumpStart model catalog within the SageMaker console or deploy them programmatically using the SageMaker Python SDK. This integration simplifies the process of adopting advanced AI solutions for a wide range of enterprise applications.

AWS customers can now leverage these foundation models to enhance their AI workflows, whether for high-performance image generation or unified multimodal processing. The availability of these models on SageMaker JumpStart underscores AWS's commitment to expanding its portfolio of accessible and scalable AI tools for developers and enterprises.

Original source → Deals on Clipraptor.com →