OpenAI is about to flip the script once more. This summer, the company plans to release its first open language model in years.
Back in 2019, OpenAI dropped GPT-2 and gave the world a taste of what AI could do. Fast forward to 2025, and the company is once again stepping into the open-source space but with bigger goals and bolder ambitions.
The upcoming model is said to be a “reasoning” model, similar to its powerful “o-series” models.
But unlike past offerings, this one will be open for developers to explore, build on, and even use commercially, without jumping through hoops.
This is an important development because many current open models from other companies, like Meta’s Llama or Google’s Gemma, still come with strings attached.
What Makes This Model Different?
According to insiders, Aidan Clark, OpenAI’s VP of Research, is heading up development. The model is still in early stages, but it’s already turning heads in the developer community.
Here’s what sets it apart:
- Truly Open License: Expect minimal usage restrictions.
- High Compatibility: Built to run on high-end consumer hardware, not just data centers.
- Switchable Reasoning: Developers might be able to toggle reasoning features on or off to balance speed and accuracy.
Why Is OpenAI Making This Move Now?
Pressure. Other AI labs, especially Chinese startups like DeepSeek, are racing ahead with fully open models. These models are freely available to the community, encouraging innovation, collaboration, and rapid iteration.
The payoff is immense as DeepSeek has earned global attention. Meta also says Llama models have been downloaded over a billion times. OpenAI, long criticized for keeping its best models behind closed doors, seems ready to give in.
Sam Altman’s Statement
OpenAI CEO Sam Altman admitted the company’s open-source strategy might have missed the mark.
“We’ve been on the wrong side of history” Altman said. That’s a pretty big admission from a company that’s dominated the AI scene.
OpenAI plans to red-team the model extensively. It will put it through rigorous safety tests and release a detailed model card, a document that explains how the model performs, what it can do, and how safe it is to use.
Altman also noted the model would go through OpenAI’s internal preparedness framework, the same process used for its most advanced models.
