The official Q4_K_S gguf is quite good and has very good 35 tps generation on a M1 mac studio. Should be much faster on recent Macs, especially M5.

It's hard to take it seriously due to its name and crappy website, I wish that they rebrand someday.

Everything looked meh so far, until entered the visual recognition and reasoning chapter. That was very impressive. Very cool to see more vendors pushing towards models understanding images instead of just performing OCR.

The pricing in their subscription plan also look quite generous. https://platform.stepfun.com/step-plan

I just wish they put a little bit of energy on making their service usable for non-chinese users. At the moment, their website assumes I speak chinese and when I manually click on "en", the website dynamically translates parts into english, I experience this as half baked solution.

Stepfun 3.5 was my daily driver, and I’ve been blown away at how good it is (even vs Qwen3.6 27B). So happy 3.7 came out!