Five new Qwen models for coding agents and efficient reasoning are now available in Amazon SageMaker JumpStart

This article announces the availability of five new Qwen models in Amazon SageMaker JumpStart, offering specialized capabilities for coding agents and efficient reasoning.

• Qwen3-Coder-Next excels at long-horizon reasoning and complex tool use for coding agents
• Qwen3-30B-A3B supports switching between thinking and non-thinking modes for general-purpose tasks
• Qwen3-30B-A3B-Thinking-2507 improves performance on complex math, science, and coding reasoning
• Qwen3-Coder-30B-A3B-Instruct designed for agentic coding with repo-scale context understanding
• Qwen3.5-4B supports vision-language training and 201 languages for lightweight deployments
• Models deployable via SageMaker Studio or Python SDK with minimal configuration

These Qwen models expand SageMaker's foundation model portfolio, enabling customers to build sophisticated AI applications for diverse enterprise use cases on AWS infrastructure.