Qwen3.6-35B-A3B-NVFP4 and Wan2.1-T2V-1.3B-Diffusers models now available on Amazon SageMaker JumpStart

Amazon SageMaker JumpStart now hosts NVIDIA’s Qwen3.6-35B-A3B-NVFP4 and Alibaba’s Wan2.1-T2V-1.3B-Diffusers, broadening the selection of foundation models available to AWS customers. The Qwen model is tailored for agentic coding and long-context reasoning, while the Wan model focuses on efficient text-to-video generation. Both are designed to run on AWS infrastructure with high performance and scalability.

Qwen3.6-35B-A3B-NVFP4 is a NVIDIA-quantized variant of Alibaba’s Qwen3.6-35B-A3B, optimized for agentic coding and multimodal reasoning. It uses a Mixture-of-Experts architecture with 35B total parameters but only 3B active per token, supporting a 262K-token context window that can be extended to about 1M tokens with YaRN scaling. Quantized to NVFP4, it maintains reasoning across turns and supports tool calling for multi-step agent workflows.

Wan2.1-T2V-1.3B-Diffusers is a lightweight text-to-video model built on a diffusion transformer with a novel Video VAE. The 1.3B-parameter model generates high-quality, physics-consistent 5-second 480p video clips from text prompts and runs on consumer-grade hardware with just 8.19 GB of VRAM. On an RTX 4090, it produces a video in roughly 4 minutes, making it one of the most accessible open-source video generation models available.

Customers can deploy either model through SageMaker JumpStart with minimal setup, either via the SageMaker console or the SageMaker Python SDK. The models are listed in the SageMaker JumpStart model catalog, enabling quick integration into AWS accounts for targeted AI applications. Documentation provides guidance on deployment and usage for foundation models in SageMaker JumpStart.