OpenAI has launched a new artificial intelligence model called GPT-OSS, making it available for free to developers, researchers, and companies around the world.
The new model comes in two versions a large 120-billion-parameter model and a smaller 20-billion-parameter version. The 20b model can run on a standard laptop with 16GB of RAM, while the 120b model requires a more powerful setup, such as an 80GB GPU.
The models mimic human reasoning and perform complex tasks, are now live on platforms such as Hugging Face, Amazon Bedrock, Groq’s inference cloud, and Saudi Arabia’s Humain AI.
What this means
Unlike many of OpenAI’s previous models, these two are classified as open-weight systems. This means developers will be able to access and use the models’ underlying parameters referred to as weights for their own AI development.
However, OpenAI is not releasing the datasets used to train the models, which means they don’t meet the full definition of open source.
OpenAI noted that it has not released an open model since 2019, when it launched GPT-2, the early predecessor of the technology behind ChatGPT.
Comparison to other AI models
The release comes months after Chinese company DeepSeek attracted global attention by launching its own open AI model, DeepSeek-V2. Meta has also made strides in open-weight models with its LLaMA series.
OpenAI’s move is seen as part of a broader shift in the industry, with increased pressure from governments and users calling for more transparency in AI development.
The U.S. government has also weighed in. In its AI Action Plan under President Trump, the White House encouraged more open AI models that can become global standards in academic and commercial applications.
Security concerns
Some critics argue that open-weight or open-source models pose security risks. OpenAI acknowledged this in July when it delayed the release to conduct more safety tests.
“We planned to launch our open-weight model next week.
“We are delaying it; we need time to run additional safety tests and review high-risk areas. we are not yet sure how long it will take us. While we trust the community will build great things with this model, once weights are out, they can’t be pulled back. this is new for us, and we want to get it right.
“Sorry to be the bearer of bad news; we are working super hard!” Altman said in a post on X
OpenAI intends for the new open models to be used by individuals, companies and governments that want to tweak and run AI systems on their own hardware and services. According to the company, Orange SA and Snowflake Inc. have already started testing the models.
According to a Bloomberg report, during a briefing with reporters this week, OpenAI said it hopes to receive feedback from users that it can then review to decide what kind of open model to release in the future.
OpenAI users can experiment with the new models on its open model playground. It also provided guides for those interested in using the models through different ecosystem providers or fine-tuning them for specific task.
