How does Qwen 3.8 27B compare on low thinking mode to the older 3.6 models?

Since we know Qwen 3.8 27B thinks quite long, but gives at least a good one-shot result where you can leave it to do everything on it own, how does it compare to the older series of models for very simple tasks where you don't want to think so long? The only fine-tune of Qwen 3.6 I genuinely enjoyed was the ThinkingCap fine tune by BottleCap. it seems to think equally as long as the base model on complex tasks, but on simple tasks it thinks shorter. Does it still make sense for me to run this older model when 3.8 27B exists with the low thinking mode?