OpenAI is preparing to release a powerful “open-weight” language model later this year, marking a significant shift from its previous closed models.
OpenAI’sCEO Sam Altman announced the move on March 31, revealing that developers will soon be able to run the model on their own hardware, a decision aimed at gathering feedback and making it “maximally useful.”
The new open-weight model, set to launch in the coming months, will be the first release of its kind since GPT-2 in 2019. Although not entirely open-source, this model will allow anyone to use, modify, and deploy it. The model is designed for greater flexibility, allowing large companies and governments to operate it on their own terms, especially in environments where data privacy and security are critical.
To ensure the model meets developer needs, OpenAI plans to host events starting in San Francisco, followed by sessions in Europe and the Asia-Pacific region. Altman expressed excitement about the potential applications developers will create and how this open-weight version could be adapted post-release.
This announcement places OpenAI in direct competition with other major AI players. Rival companies like DeepSeek, Alibaba with its open-source AI model, and Google’s Gemini 2.5 have introduced cost-effective alternatives in the AI arms race. As the AI landscape heats up, OpenAI’s shift toward open-weight models promises to fuel innovation across industries, while maintaining control over post-release modifications.
Kosta Gushterov
Reporter at Coindoo
Kosta has been a part of the team since 2021 and has solidified his position with a thirst for knowledge, incredible dedication to his work and a “detective-like” mindset. He not only covers a wide range of trending topics, he also creates reviews, PR articles and educational content. His work has also been referenced by other news outlets.
Source: https://coindoo.com/openai-to-launch-open-weight-language-model-for-developers/