Product LaunchSimon Willison
Qwen 3.8 27B is excellent, but it defaults to wildly overthinking things
#qwen#llm#ai#model#performance
English
Qwen 3.8 27B, a new model from Alibaba, shows impressive performance but defaults to excessive reasoning, leading to inefficiencies in generating outputs. Users are advised to adjust the reasoning settings to improve efficiency, as the model can produce high-quality results when configured correctly.
中文
阿里巴巴的新模型Qwen 3.8 27B表现出色，但默认的过度推理设置导致生成输出效率低下。建议用户调整推理设置，以提高效率，因为在正确配置时，该模型可以产生高质量的结果。