I openly hope the chinese labs distilling them into open weights win.

Yup. I'm done with US companies. Let's go China !

Tried Xiaomi MiMo v2.5 via opencode today. Since Sonnet 5's release week, Sonnet 4.6 has been feeling like a vegetable, with Sonnet 5 itself being only a little better. MiMo on the other hand feels like Sonnet 4.6 did up until very recently. Absolutely impressive.

In some ways, more impressive than GPT 5.5 with high(!) thinking. GPT says quite some nonsense from time to time; didn't see any sign of this in MiMo so far, which is a pretty wild difference.