Skip to content
RCreddit.com·

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

AI summary

A user is asking for a comparison between Qwen 3.8 27B in 'low thinking mode' and older Qwen 3.6 models, specifically for simple tasks. They note that Qwen 3.8 27B typically 'thinks quite long' but delivers good one-shot results. The user previously enjoyed the 'ThinkingCap' fine-tune of Qwen 3.6, which thought shorter on simple tasks, and is wondering if it's still worthwhile to use this older model given the new 3.8 27B's low thinking mode.

Why this one

This post uniquely compares Qwen 3.8 27B's new 'low thinking mode' with older 3.6 models, specifically addressing performance on simple tasks, unlike general benchmarks.

Time & source

Times shown in UTC

Display time zone: UTC

Local time zone unavailable; showing UTC.

PublishedOffset at this time: UTC+0Sep 13, 2026, 09:43 UTC

IngestedOffset at this time: UTC+0Sep 13, 2026, 15:01 UTC

Published
Sep 13, 2026, 09:43
Ingested
Sep 13, 2026, 15:01
Source type
Dev community
Tier
Community
Source status
Healthy

Tier is a per-source editorial setting, not a per-item score.

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?

Source·reddit.com