Openai says it intends to release its first “open” language model since GPT ‑ 2 “in the coming months.”
This is what it is, according to a feedback form the company published on its website on Monday. The forms Openai invites developers, researchers, and (members of the broader community) to fill out include questions such as, “What do you want to see in Openai’s openweight model?” “What kind of open models have you used in the past?”
“We are excited to work with developers, researchers and the broader community to gather input and make this model as useful as possible,” Openai wrote on its website. “If you are interested in participating in a feedback session with the Openai team, please let us know (in the form) below.”
Openai plans to host developer events to collect feedback and will collect demo prototypes of the model in the future. The first developer event will be held in San Francisco within weeks, followed by sessions in Europe and Asia Pacific.
This year I’m releasing a model that can run on my own hardware https://t.co/0ji9oeznyr
– March 31, 2025, Stephen Heidel (@stevenheidel)
Openai is facing increased pressure from rivals such as the Chinese AI Lab Deepseek, which took a “open” approach to launch the model. In contrast to Openai’s strategy, these “open” competitors make the model available to the AI community for experimentation and, in some cases, commercialization.
It has proven to be a hugely successful strategy for some outfits. Meta, which invested heavily in the Lama family of open AI models, said early March that Llamas had won over 1 billion downloads. Meanwhile, Deepseek quickly garnered a large global user base and attracted the attention of domestic investors.
In a recent Reddit Q&A, Openai CEO Sam Altman said he believes Openai is on the wrong side of history when it comes to sourcing technology.
“I know another open source strategy (I personally think I need it),” says Altman. “Not everyone at Openai shares this view, nor is it a top priority at the moment (…) we will create a better model (to the future), but we will not maintain a more lead than last year.”
Altman extended Openai’s open model plan with an X post Monday afternoon, saying Openai’s upcoming open model has a “inference” feature along Openai’s O3-Mini line.
“(b) Before release, we evaluate this model according to the preparation framework, like with other models,” says Altman. “(a) we do some extra work considering we know that this model will change after release (…) (W) we see what developers will build and how they use it where big companies and governments prefer to run the model themselves.”
An excerpt from an upcoming book by Keach Hagey of the Wall Street Journal Reporter claims that Altman misinterpreted Openai executives about the safety review of the model in November 2023 before his brief expulsion.