Amazon is significantly expanding its reach in the artificial intelligence landscape by embracing open-source AI models, particularly those from OpenAI. This move provides developers and businesses with greater access to powerful AI technologies, fostering innovation and offering more control over their AI application development.

Availability of OpenAI Models on AWS Platforms

Amazon Web Services (AWS) has announced the availability of OpenAI's open-weight models, gpt-oss-120b and gpt-oss-20b, on Amazon Bedrock and Amazon SageMaker JumpStart. This marks the first time AWS has offered products from OpenAI, effectively breaking Microsoft's exclusive hold on cloud distribution of these models. These models are designed for text generation and reasoning tasks and are well-suited for coding, scientific analysis, and mathematical reasoning.

The open-weight status of these models means that the underlying parameters, or weights, that determine their behavior are freely available for download. This allows developers and companies to customize, fine-tune, and deploy the models themselves, offering greater flexibility and control. These models are available under the Apache 2.0 free software license, meaning they are not subject to the exclusive provisions of OpenAI's agreement with Microsoft.

Key Features and Capabilities

Both the gpt-oss-120b and gpt-oss-20b models feature a 128K context window and adjustable reasoning levels (low/medium/high) to match specific use case requirements. They also support external tool integration and can be used in agentic workflows through frameworks like Strands Agents. Furthermore, these models offer full chain-of-thought output capabilities, providing detailed visibility into the model's reasoning process.

Benefits for Developers and Businesses

By offering these open-weight models, Amazon aims to empower developers and organizations with more powerful AI technologies. Customers can leverage the models' advanced reasoning capabilities for various applications, including agentic workflows, coding, scientific analysis, and mathematical problem-solving. The availability of these models on AWS allows customers to build generative AI applications quickly and easily.

Amazon Bedrock AgentCore allows customers to deploy and operate highly effective agents using OpenAI models, providing the scale and security required for production applications. The models also give users the freedom to modify, adapt, and customize them for their specific business needs.

Amazon's Broader AI Strategy

This move aligns with Amazon's broader AI strategy, which includes investing in various AI initiatives and partnerships. Amazon has invested $8 billion in Anthropic, the maker of the Claude chatbot, and offers Anthropic's models on the Bedrock platform. AWS also provides tools and frameworks such as Kiro, Amazon Q Developer, and Strands Agents to help customers build and deploy AI agents.

Amazon is also focused on providing AI skills training to help businesses execute their AI strategies. The company has trained two million people globally with free generative AI skills as part of Amazon's AI Ready commitment.

Security Considerations

While embracing open-source AI offers numerous benefits, security remains a critical concern. A recent incident involving Amazon Q Developer, where a hacker infiltrated the tool with malicious code, highlights the vulnerabilities in AI-driven software development. Amazon has emphasized the importance of security in building and deploying AI agents for enterprises.

Availability and Pricing

The new OpenAI open-weight models are currently available in Amazon Bedrock in the US West (Oregon) AWS Region, while Amazon SageMaker JumpStart supports these models in the US East (Ohio, N. Virginia) and Asia Pacific (Mumbai, Tokyo) AWS Regions. AWS has stated that OpenAI's larger open-weight model offers enterprises significantly better value for the price compared to competing models.

By making OpenAI's open-weight models accessible on its platforms, Amazon is taking a significant step toward democratizing AI and empowering developers and businesses to build innovative AI applications.