I Tested Alibaba Qwen3.6–35B-A3B vs Google Gemma 4 26B A4B — The Smaller-Active Model Won Coding…

Five days ago, Alibaba’s Qwen team quietly dropped a new open-weight MoE model called Qwen3.6–35B-A3B — 35 billion total parameters, but…