I'm a believer in StepFun Westerners mostly don't dig their work, but I think 3.5 Flash was a genuine breakthrough for open models at the time, in STEM and in FAST agentic behavior. Not sure why 3.6-3.7 don't make GLM/Kimi level progress, but I expect they'll leave a bigger mark.
Xiuyu Li@sheriyuo
Personal Update: I've joined @StepFun_ai as a Researcher, working on LLM post-training.
Many thanks to @CyouSakura and @MikaStars39 for their support.
Together, we'll take another step toward the frontier.
6:34 AM · Jun 24, 2026 · 3.9K Views
