Developers have the opportunity to weigh in on OpenAI’s latest project. On March 31, the AI giant published applications for feedback sessions on an upcoming open language model, the second such model since OpenAI’s LLMs went private after GPT-2. It will be released “in the coming months,” according to a post on X by OpenAI CEO Sam Altman.
Open-weight models can be modified by their users. In particular, “weights” in generative AI refers to the connections between digital neurons; changing them might change which concepts the model associates together or the emphasis put on those associations.
In Altman’s post, he said the model currently in development will be a “reasoning” model like OpenAI o1.
“We’ve been thinking about this for a long time but other priorities took precedence. now it feels important to do,” Altman said.
Developers can comment on what they want from an open model
Developers in San Francisco will be able to attend an OpenAI event “to gather feedback and later play with early prototypes” in a couple of weeks, he said. OpenAI will hold similar events in Europe and APAC; Altman did not disclose specific locations for those events.
“Before release, we will evaluate this model according out our preparedness framework, like we would for any other model. and we will do extra work given that we know this model will be modified post-release,” Altman wrote.
OpenAI’s feedback form
OpenAI is soliciting feedback about the open model from developers, as well as researchers and anyone in the broader community. The application on OpenAI’s site asks interested developers to provide links to their GitHub profiles and social or professional profiles. It also asks what open models the applicant has used in the past, what they would want to see in an open-weight model from OpenAI, and what they would use it for.
“We’re excited to collaborate with developers, researchers, and the broader community to gather inputs and make this model as useful as possible,” OpenAI wrote.
SEE: xAI has acquired social media platform X for $45 billion, a consolidation of Elon Musk’s AI startup and the platform formally known as Twitter.
The definition of open source can be murky. One organization, the Open Source Initiative, defines open source AI as models that can be used for any purpose, modified in any way, studied to see how they arrived at their output, and shared with others.
More must-read AI coverage
Open models are a point of differentiation among AI giants
As TechCrunch pointed out, OpenAI’s competitors have more recent examples of success with open models. Developers can download Llama AI model weights from Meta; DeepSeek opened its R1 and V3 models, including some of the model code.
OpenAI may be banking on the idea that letting developers run the model themselves will give them a competitive edge in the increasingly crowded AI market.