LocalLLaMA

Qwen 3.6 35B A3B vs Qwen 3.5 122B A10B

Does anyone else have the same experience comparing these two – for me 3.5 122B outperforms 3.6 by a large margin. 3.6 gets lost as long as the task requires a couple of more steps. I'm asking because I got the impression that it overperforms in so…