Amazon Web Services Takes a Major Step in AI with New Models on Amazon Bedrock

Amazon Web Services Takes a Major Step in AI with New Models on Amazon Bedrock

Amazon Web Services (AWS) has taken a major step in the AI landscape by making OpenAI’s open-weight models, Anthropic’s Claude 4 series, and Alibaba’s Qwen3 family available on the Amazon Bedrock platform.

What Are Amazon Bedrock and SageMaker AI?

Amazon Bedrock is a fully managed service for building AI-powered applications. Through secure APIs, developers can access a wide range of foundation models and build their own applications with enterprise-grade security, scalability, and model diversity.

Amazon SageMaker AI is a comprehensive platform for training, deploying, and managing machine learning models. Data scientists and developers can handle every stage of the ML lifecycle within SageMaker. Through SageMaker JumpStart, users gain access to dozens of pre-built language models.

OpenAI Open-Weight Models: The gpt-oss Series

OpenAI is offering its open-weight models on AWS for the first time. The gpt-oss-120b and gpt-oss-20b models deliver strong performance in agentic workflows, coding, and scientific analysis.
These models provide up to 10× better cost-performance compared to comparable Gemini models and support 128K context windows, enabling efficient processing of long documents.

Anthropic Claude 4: Hybrid Reasoning

Anthropic’s Claude Opus 4.1 and Claude Sonnet 4 introduce a hybrid reasoning system that can switch between instant responses and deeper analytical thinking.
Claude Opus 4.1 is recognized as an industry leader in coding, achieving a 74.5% success rate on the SWE-bench benchmark. Claude Sonnet 4, on the other hand, offers an optimized cost-performance balance for high-volume use cases.

Qwen3 and DeepSeek-V3.1: Multilingual AI Models

Alibaba’s Qwen3 family offers four different models with strong performance in Chinese and English. While the Qwen3-Coder series focuses on software engineering tasks, the general-purpose models include both “thinking” and “non-thinking” reasoning modes.

DeepSeek-V3.1, with 685 billion parameters, delivers hybrid reasoning capabilities. By activating only the required knowledge segments, the model achieves cost efficiency, while excelling in software development and mathematical analysis.

Security and Enterprise Capabilities

Amazon Bedrock provides enterprise-grade security, including data encryption, strict access controls, and Amazon Bedrock Guardrails, which can block up to 88% of harmful content. AWS does not share customer data with model providers, ensuring data privacy and compliance.

These developments unlock new opportunities for AI agents and automated workflows across healthcare, finance, and manufacturing. The transparency and customization advantages of open-weight models accelerate AI innovation while significantly improving cost-performance ratios.

For more information, you can visit the official Amazon Bedrock web page, or contact us at ai@empa.com
 for project-specific AI support.