OpenAI is about to loosen its grip, unleashing its first “open-weight” language model since the GPT-2 days of 2019. CEO Sam Altman announced via X on March 31 that the company will drop a powerful, “reasoning” AI model into the wild soon.
This move towards a more accessible model is a departure from OpenAI’s recent strategy of tightly controlled releases like GPT-3 and GPT-4. Altman specified that they are actively seeking feedback to optimize the model for “maximal usefulness.”
To gather input, OpenAI will kick off developer events in San Francisco, extending to Europe and the Asia-Pacific region. Altman emphasized that this “open-weight” model, distinct from a fully “open-source” one, will empower developers, large corporations, and governments to operate the AI on their own systems and will allow users to use, download, modify and deploy the resultant model.