OpenAI has delayed the release of its planned open-weight language model for the second time in less than a month. Originally expected to launch in July, the release has now been pushed back with no set date as the company continues internal safety reviews.
CEO Sam Altman announced the delay on X (formerly Twitter), saying the model still needs more testing before release. He stressed that once an open-weight model is published, it cannot be taken back, so they must proceed carefully.
we planned to launch our open-weight model next week.
we are delaying it; we need time to run additional safety tests and review high-risk areas. we are not yet sure how long it will take us.
while we trust the community will build great things with this model, once weights are out, they can’t be pulled back. this is new for us and we want to get it right.
— Sam Altman (@sama)
Although this model isn’t meant to match the power of GPT-4 or GPT-4o, it would still be OpenAI’s first time releasing a model with open weights, giving developers the freedom to use and modify it on their own systems. The delay happens at a time when rivals like Meta, Mistral, and China’s Moonshot AI are moving forward with their own open models. Moonshot, for instance, has recently introduced Kimi K2, a massive trillion-parameter model that has drawn wide international attention.
The move has caused mixed reactions in the AI community. Some researchers support OpenAI’s caution, highlighting real safety risks. Others worry delays could slow innovation and reduce transparency, especially as rivals speed up open-source efforts.
OpenAI has not set a new release date but said it will provide updates after completing safety evaluations.
