OpenAI’s CEO, Sam Altman, announced on Friday that the release of the company’s open model is being postponed indefinitely for additional safety testing. Originally scheduled for next week, the open model’s launch has been delayed to ensure thorough safety checks and review of high-risk areas. Altman emphasized the importance of getting it right before releasing the model to the community, as once the weights are out, they cannot be retracted. This decision reflects OpenAI’s commitment to delivering a secure and reliable product.
OpenAI had initially planned to unveil its open model this summer, alongside the highly anticipated release of GPT-5 from the creators of ChatGPT. Unlike GPT-5, OpenAI’s open model will be freely accessible for developers to download and utilize locally. This delay means that developers will have to wait a bit longer to experience the innovative capabilities of OpenAI’s latest offering. The competitive landscape of open AI models has also intensified with the recent launch of Moonshot AI’s Kimi K2, a one trillion parameter model that surpasses OpenAI’s GPT-4.1 in various benchmarks.
Altman’s announcement of the delay comes after the company achieved a significant milestone, prompting the need for additional time to ensure a high-quality release. Aidan Clark, OpenAI’s VP of research leading the open model team, emphasized the exceptional capabilities of the model but noted the importance of meeting the company’s standards for an open-source release. The article also mentions discussions within OpenAI about potential features for the open model, such as connectivity to the company’s cloud-hosted AI models for complex queries.