OpenAI has officially made its open weight models available on AWS, allowing millions of Amazon Web Services customers to use OpenAI’s advanced models gpt oss 120b and gpt oss 20b through Amazon Bedrock and Amazon SageMaker AI. This major move puts powerful generative AI tools directly into the hands of developers, researchers, and enterprises already building in the AWS ecosystem.
Starting today, AWS customers can access these OpenAI models for building and deploying generative AI applications with use cases like agent-based systems, coding assistants, scientific analysis, and solving mathematical problems. The new models, especially gpt oss 120b, are optimized for high performance and cost efficiency 3 times more cost effective than Gemini, 5x than DeepSeek R1, and 2x than OpenAI’s own o4 model.
This launch expands AWS’s model catalog, which already includes more than 100 models, and adds OpenAI to its list of open weight providers, alongside Meta, Mistral AI, and DeepSeek.
What These Models Offer:
- Advanced Reasoning: Designed for agentic AI, they support step by step reasoning, chain-of-thought prompts, and complex decision trees.
- Enterprise-Grade Security: Integrated with Bedrock tools like Guardrails, which filters up to 88% of harmful content.
- 128K Context Window: Supports longer text input and conversations, ideal for documents, transcripts, and academic content.
- Instruction Following & Tool Use: Models can browse the web, use code interpreters, and follow custom instructions.
Swami Sivasubramanian, AWS VP for Agentic AI, announced the news during the AWS Summit in New York 2025, where AWS also introduced AgentCore, a toolkit to help developers build and deploy AI agents faster. The announcement included a $100 million investment to support the growth of agentic AI development.
Integration Details:
- Amazon Bedrock: Customers can build and scale with full model management, native security, and integrations like Custom Model Import and Knowledge Bases.
- Amazon SageMaker JumpStart: Enables model evaluation, fine-tuning, and deployment in a streamlined environment.
Dmitry Pimenov, product lead at OpenAI, said, “Our open weight models help developers from solo builders to large enterprise teams unlock new possibilities across industries and use cases.”
Atul Deo, AWS product director, added that this collaboration marks a “natural progression” in AWS’s plan to democratize advanced AI tools globally.
Who’s Already Using It:
Organizations like Pfizer, LexisNexis, Rocket Companies, GoDaddy, DoorDash, and Siemens are already relying on Amazon Bedrock to deploy generative AI at scale. With OpenAI now on board, more developers will have access to best in class AI performance combined with the infrastructure power of AWS.
The open weight models offer flexibility and freedom, allowing businesses to customize and control their AI solutions without being locked into closed systems. And with AWS’s expansive ecosystem, the tools are ready for production level use cases.
This move is part of AWS’s broader strategy to support the future of generative AI, and OpenAI’s inclusion strengthens their position as a leader in accessible, safe, and scalable AI deployment across industries.
