Qwen 3.8 and GLM 5.3 are my workhorses right now. Qwen for simpler stuff because it is wicked fast using 'ninfer', GLM 5.3 because it is very good a more complex stuff and less likely to get stuck so I can let it run for much longer without intervention.