OpenAI releases pair of open reasoning models

Microsoft-backed (NASDAQ:MSFT) OpenAI launched a pair of open weight models today, which can compete with other open models such as Google’s (NASDAQ:GOOG)(NASDAQ:GOOGL) Gemini CLI and DeepSeek’s (DEEPSEEK) R1 model.

OpenAI’s new open weight foundation models are called gpt-oss-120b and gpt-oss-20b. The 120b is a large open model designed to run in data centers or on a high-end desktop or laptop powered by an Nvidia (NVDA) H100 GPU or better. The 20b is a medium-sized model that can run on most desktops and laptops.

“Running in Amazon Bedrock (NASDAQ:AMZN), the larger of the two open weight models is 3x more price-performant than the comparable Gemini model, 5x more than DeepSeek-R1, and 2x more than the comparable OpenAI o4 model,” according to Amazon Web Services.

“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,” said AWS director of product Atul Deo.

The models can also be downloaded through Hugging Face or viewed through GitHub.

The models are supported by the Apache 2.0 license, which allows developers to build without the worry of copyright or patent issues, even for commercial deployments, OpenAI said.

“Our open weight models help developers—from solo builders to large enterprise teams—unlock new possibilities across industries and use cases,” said Dmitry Pimenov, product lead at OpenAI.