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

Amazon Web Services has expanded its SageMaker JumpStart platform with five new Qwen foundation models designed for specialized AI applications including coding agents and advanced reasoning tasks. The newly available models include Qwen3-Coder-Next for powering coding agents with long-horizon reasoning and error recovery capabilities, Qwen3-30B-A3B featuring switchable thinking modes for general assistant tasks, and Qwen3-30B-A3B-Thinking-2507 optimized for complex mathematical and scientific reasoning with enhanced long-context understanding. The release also includes Qwen3-Coder-30B-A3B-Instruct, which supports agentic coding workflows with custom function calling and repository-scale context processing, and Qwen3.5-4B, a lightweight multimodal model supporting 201 languages and unified vision-language training. AWS customers can deploy these models directly through SageMaker Studio's interface or programmatically via the SageMaker Python SDK, providing immediate access to specialized AI capabilities without requiring extensive machine learning infrastructure setup.