4 new Qwen models for multimodal reasoning, agentic coding, and multilingual applications are now available in Amazon SageMaker JumpStart

Posted on: May 4, 2026

Today, we are excited to announce the availability of four new Qwen models in Amazon SageMaker JumpStart: Qwen3.5-27B-FP8, Qwen3.6-35B-A3B, Qwen3.5-0.8B, and Qwen3.5-2B.

These models address different AI application needs with specialized capabilities:

• Qwen3.5-27B-FP8 – A multimodal vision-language model for reasoning over images, video, and text. Designed for applications such as agentic tool use, coding assistance, complex mathematical reasoning, multilingual communication in over 200 languages, and long-context processing with support for up to 1 million tokens.
• Qwen3.6-35B-A3B – An efficient Mixture-of-Experts model with 3 billion active parameters optimized for agentic coding workflows. Suited for tasks including frontend development, repository-level code reasoning, multi-step agent interactions, and coding copilot applications.
• Qwen3.5-0.8B – A compact multimodal model designed for rapid prototyping, fine-tuning, on-device inference, and edge deployments. Supports multilingual capabilities and multimodal understanding at a minimal compute footprint.
• Qwen3.5-2B – A lightweight multimodal model for prototyping, fine-tuning, and moderate-compute deployments. Handles multilingual text generation, visual understanding, and conversational AI tasks efficiently.

All four models are available today through Amazon SageMaker JumpStart. You can deploy them with a few clicks in Amazon SageMaker Studio or programmatically using the SageMaker Python SDK.