OpenAI has released an open-source pretrained language model for the first time since GPT-2 in 2019.

BlockBeats News, April 1st, according to Cointelegraph, OpenAI founder Sam Altman announced that in the coming months, they will release a new open-source weighted language model with reasoning capabilities. Altman first seeks feedback on "how to maximize its utility." An open-source weighted language model refers to a model that can be publicly used, downloaded, modified, or deployed. While not as open as full open-source models, this marks a significant shift from the fully closed GPT-3 and GPT-4. In February 2019, OpenAI partially open-sourced the GPT-2 model, and by November of the same year, it was fully open-sourced. On February 12th of this year, Sam Altman stated that the company plans to launch GPT-4.5 and GPT-5 in the coming weeks or months.