OpenAI delays its open model release indefinitely for extra safety testing. The company had set the launch for next week but pushed it back again on Friday. Sam Altman said the move is about making sure the model is safe before letting it loose.
Altman took to X to explain the hold-up.
Sam Altman stated: 
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.
This open model was supposed to be OpenAI’s first in years to be freely downloadable and run locally by developers. It’s expected to have reasoning powers on par with OpenAI’s “o-series” models and compete with other open AI models. The delay means developers will have to wait longer to test it out.
Meanwhile, the AI landscape is heating up. Chinese startup Moonshot AI just launched Kimi K2, a one trillion parameter open model that beats OpenAI’s GPT-4.1 on some coding benchmarks.
OpenAI’s VP of research Aidan Clark, who leads the open model team, also shared thoughts on X:
Capability wise, we think the model is phenomenal — but our bar for an open source model is high and we think we need some more time to make sure we’re releasing a model we’re proud of along every axis.
Back in June, Altman said OpenAI had made something “unexpected and quite amazing” with this open model but didn’t say more.
There’s talk OpenAI might let this open model connect to its cloud AI for complex requests. It’s unclear if that feature will be part of the final release.
For now, developers and the AI community will have to wait longer to dig into this highly anticipated open AI milestone.
