AnalysisAI ModelsAugust 16, 2026

Qwen 3.8 27B defaults to overthinking, says Simon Willison

Qwen 3.8 27B, an Apache 2 licensed 27B vision-capable LLM from Alibaba, defaults to xhigh reasoning effort, causing spectacular overthinking. Simon Willison found it used 22,276 reasoning tokens and took 21 minutes to generate a simple SVG.

1 source

More stories today

Open the live feed