OpenAI has released two new open-weight language models — gpt-oss-120b and gpt-oss-20b — in a strategic move to reassert its dominance in the fast-evolving world of artificial intelligence.
The announcement, made on Tuesday, marks the company’s first open-weight model release since GPT-2 in 2019 and comes amid a heated contest among global AI developers, notably Meta and China’s DeepSeek, for leadership in the open-weight and open-source model space.
Unlike open-source models that make training data and methodology public, open-weight models provide access to trained parameters (weights), allowing developers to fine-tune and deploy them without access to the original training data. OpenAI emphasized that these models are designed to run efficiently on consumer-grade devices.
Register for Tekedia Mini-MBA edition 19 (Feb 9 – May 2, 2026): big discounts for early bird.
Tekedia AI in Business Masterclass opens registrations.
Join Tekedia Capital Syndicate and co-invest in great global startups.
Register for Tekedia AI Lab: From Technical Design to Deployment (next edition begins Jan 24 2026).
The larger model, gpt-oss-120b, runs on a single GPU, while the smaller gpt-oss-20b is optimized for laptops and PCs. Both models are built to perform on par with OpenAI’s proprietary o3-mini and o4-mini reasoning models, particularly excelling in advanced reasoning, coding, competition mathematics, and health-related queries.
“One of the things that is unique about open models is that people can run them locally. People can run them behind their own firewall, on their own infrastructure,” said Greg Brockman, co-founder of OpenAI, during a press briefing.
The models were trained on a text-only dataset with a heavy focus on science, math, and programming domains critical to performance in next-gen AI tools.
Adding to the momentum, Amazon Web Services (AWS) has onboarded OpenAI’s new models onto its Bedrock generative AI marketplace. This makes them accessible to a broader developer base, marking the first time any OpenAI model has been hosted on Bedrock. Atul Deo, Director of Product at AWS Bedrock, said the partnership gives customers powerful open-weight options.
“OpenAI has been developing great models and we believe that these models are going to be great open-source options, or open-weight model options for customers,” Deo said in an interview.
The rollout also coincides with OpenAI’s ongoing $40 billion funding round led by Japan’s SoftBank Group, which could raise its valuation to an eye-popping $300 billion. The timing of the release and the scale of fundraising reflect OpenAI’s intensifying efforts to entrench its position against a growing field of global competitors.
Meta’s LLaMA (Large Language Model Meta AI) models were long seen as leaders in the open-weight space until China’s DeepSeek emerged with DeepSeek-V2 and DeepSeek-Coder, models that combined raw performance with cost efficiency. DeepSeek’s releases forced a recalibration in the market, especially as Meta’s anticipated LLaMA 4 suffered repeated delays.
Meanwhile, OpenAI’s move is also seen as a counter to the growing shift in developer preference towards models that offer greater transparency and customizability. This demand for openness has propelled DeepSeek and Mistral — a French startup — into the global spotlight. Mistral, for instance, recently secured a €600 million funding round to expand its suite of open-weight models and enterprise AI tools. Likewise, China’s Zhipu AI, supported by Alibaba and Tencent, has been pushing out its GLM family of models with a focus on multilingual capabilities and commercial deployment.
Analysts believe OpenAI’s reentry into open models is a strategic pivot that acknowledges the broader industry shift. Some note that closed models like GPT-4 are powerful but inaccessible for developers looking to embed models in private or on-prem systems. OpenAI’s new models are designed to address that gap.
However, OpenAI has not released performance benchmarks comparing its gpt-oss models with DeepSeek or Mistral’s open-weight counterparts, leaving room for speculation about whether the move is more strategic than revolutionary.
OpenAI’s re-emergence in the space may recalibrate the balance of power as the artificial intelligence industry rapidly shifts towards openness, but it faces steep competition from agile rivals in China and Europe who have already captured critical developer mindshare.