OpenAI has announced a landmark partnership with Amazon Web Services (AWS), bringing its advanced open weight foundation models directly to AWS customers. For the first time, organizations can access OpenAI’s gpt-oss-120b and gpt-oss-20b models via Amazon Bedrock and Amazon SageMaker AI, paving the way for a new era in generative AI development within the world’s most widely used cloud ecosystem.

Expanding Access to Powerful AI Models

Millions of AWS customers can now quickly deploy OpenAI’s new models, which deliver exceptional reasoning capabilities for a wide range of use cases, including agentic workflows, software coding, scientific research, and advanced mathematical problem-solving. These models significantly broaden the selection of generative AI tools available on AWS, further establishing the platform as a leader in model choice and flexibility for enterprises and developers.

Amazon Bedrock and SageMaker: Seamless Integration and Performance

OpenAI’s gpt-oss-120b, the larger of the two models being launched, is set to redefine price-performance in the market, delivering three times better cost efficiency than comparable Gemini models, five times compared to DeepSeek-R1, and double the efficiency of the OpenAI o4 model when running in Amazon Bedrock. Customers can integrate these models with Amazon Bedrock’s robust security features, like Guardrails, which helps block up to 88% of harmful content, and take advantage of tools supporting Custom Model Import, Knowledge Bases, and future customization capabilities.

Amazon SageMaker AI users will also benefit from OpenAI’s open weight models, gaining comprehensive resources for pre-training, evaluation, fine-tuning, and deployment, allowing teams to maximize the models’ capabilities across their AI projects.

Driving the Future of Agentic AI

The rollout at AWS Summit New York 2025 underscores AWS’s commitment to innovation in AI agents, with Amazon Bedrock AgentCore enabling the deployment and management of high-performing agents at scale. Supported by a recent $100 million investment to accelerate agentic AI development, AWS is making cutting-edge technology available for real-world production use, delivering the scale and security customers have come to expect.

Advanced Reasoning, Enhanced Use Cases

OpenAI’s open weight models shine in agentic AI scenarios, thanks to their adjustable reasoning levels and chain-of-thought capabilities for breaking down complex problems. Developers gain tools for instruction-following, web search, code interpretation, and more, making these models a strong fit for handling extended tasks such as customer service transcripts, technical documentation, and academic literature through their 128K token context window.

Industry Leaders Speak Out

Open weight models are an important area of innovation in the future development of generative AI technology, which is why we have invested in making AWS the best place to run them—including those launching today from OpenAI. The addition of OpenAI as our newest open weight model provider marks a natural progression in our commitment to bringing cutting-edge AI to organizations worldwide, and the unmatched size of our customer base marks a transformative shift in access to OpenAI’s advanced technology.

 Atul Deo, director of product, AWS.

Our open weight models help developers—from solo builders to large enterprise teams—unlock new possibilities across industries and use cases. Together with AWS, we’re providing powerful, flexible tools that make it easier than ever for customers to build, innovate, and scale.

Dmitry Pimenov, product lead, OpenAI.