Qwen 3.8 27B performs well but defaults to excessive reasoning intensity, causing overthinking
Original title:Qwen 3.8 27B is excellent, but it defaults to wildly overthinking things
AI summary
Alibaba's Qwen lab released Qwen 3.8 27B, a 27B-parameter vision large model under the Apache 2 license. Official benchmarks show it surpasses its predecessor Qwen 3.6 27B and the closed-source Qwen 3.7-Plus.
Key Takeaways
- Alibaba's Qwen lab released Qwen 3.
- 8 27B,a 27B-parameter vision large model under the Apache 2 license.
- Official benchmarks show it surpasses its predecessor Qwen 3.
- 6 27B and the closed-source Qwen 3.
Quick Facts
- Published
- Updated
- Source
- Simon Willison 博客
- Category
- Tutorials & Tips
- AI score
- 73
把 Qwen 3.8 27B 的推理档位成本量化成时间差异,xhigh 默认让生成同一 SVG 从 137 秒膨胀到 21 分钟,这个默认值比能力更影响本地选型。
Related entities
Source & copyright
This article is based on curated titles and AI summaries from the AIHOT public API, with attribution and links back. Original copyright remains with the original publishers and authors; this site does not mirror full texts.
How to Cite
When citing this article, keep the title, date, and permalink, and credit Orange News as the source.
Orange News: Qwen 3.8 27B performs well but defaults to excessive reasoning intensity, causing overthinking. August 17, 2026. https://new.nat6.net/en/news/cmswesk6a0r2hrovml2edcl2c/