返回全部动态
Qwen 3.8 27B评测：性能出色但默认过度思考
原标题：Qwen 3.8 27B is excellent, but it defaults to wildly overthinking things
AI 摘要
阿里Qwen实验室发布Apache 2.0许可的Qwen 3.8 27B视觉语言模型，作者Simon Willison在本地设备上测试后发现其默认推理强度为xhigh，导致模型对简单任务过度思考，生成耗时极长。作者建议用户将推理强度调低或关闭，并展示了模型在生成SVG和边界框标注方面的优秀能力。
以上摘要由 AI 生成，可能存在误差。事实请以原文为准。
正文节选
Qwen 3.8 27B is excellent, but it defaults to wildly overthinking things 16th August 2026 Friday’s big release was Qwen 3.8 27B, an Apache 2 licensed 27B parameter vision-capable LLM from Alibaba’s Qwen research lab. I’ve been looking forward to this one: 27B is an excellent size for running a model on a reasonably specced laptop, and its predecessor Qwen 3.6 27B was impressive. Qwen’s self-reported benchmarks for this model are eye-opening. They show a boost from both Qwen 3.6 27B and the close
发布时间：2026-08-17 06:00
抓取时间：2026-08-17 06:11
来源机构：Simon Willison