Step 3.7 Flash

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

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.