AWS has announced a significant partnership with OpenAI, making their open weight models available on the platform for the first time. This move expands the range of models available on AWS, allowing customers to leverage advanced reasoning capabilities in agentic workflows, coding, scientific analysis, and mathematical problem-solving.
The two new open weight foundation models from OpenAI – gpt-oss-120b and gpt-oss-20b – will make powerful AI technologies accessible to organizations worldwide. Running on Amazon Bedrock, these larger models are 3x more price-performant than comparable Gemini and DeepSeek-R1 models.
OpenAI’s open weight models offer advanced reasoning capabilities, including adjustable reasoning levels and chain-of-thought outputs that break down complex problems into logical steps. These models support instruction-following and tool use, enabling customers to reference real-time information and perform multi-step tasks.
Amazon Bedrock already empowers tens of thousands of global customers with the flexibility to build, customize, and innovate on generative AI applications. With this new partnership, OpenAI joins other open weight model providers like DeepSeek, Meta, and Mistral AI, further expanding customers’ ability to select and combine optimal models for their unique requirements.
This development reinforces AWS’s commitment to delivering innovative generative AI technologies to meet customer needs. With the addition of OpenAI’s open weight models, customers can unlock new possibilities across industries and use cases, making it easier than ever to build, innovate, and scale.
Source: https://www.aboutamazon.com/news/aws/openai-models-amazon-bedrock-sagemaker