This smaller model, called DeepSeek-R1-0528-Qwen3-8B, might not be as big or powerful as the full R1, but it still performs impressively well, especially in solving complex math problems.

A team led by the chief scientist, Ilya Sutskever, made a breakthrough earlier this year that allowed the company to build a model that could solve math problems, a report said.

If OpenAI's new model can solve grade-school math, it could pave the way for more powerful systems.