Alibaba Releases Qwen4-Max and Work360 Benchmark
Alibaba released Qwen4-Max. Its release materials showed the model generating text and assisting with tasks including preparing financial research materials, organizing legal due diligence and planning automobile production lines. The team also announced Work360, an industry benchmark for assessing models on complex professional tasks. The report said these demonstrations still need to be tested in real-world settings.
The release focuses on professional work, while Work360 is designed to evaluate models on completing complex professional tasks rather than merely answering questions. The demonstrations remain subject to real-world validation.
今天,阿里发布了 Qwen4-Max。发布材料所展示的,除了回答问题、生成文字,还有制作金融投研底稿、整理法律尽调材料、规划汽车产线。团队同时公布了Work360行业基准,评估模型完成复杂专业任务的能力。
这些演示仍需在真实场景中检验,但它们提出了一个共同问题:要让AI交付专业成果,该拿什么教它,又该如何判断它做得对不对?
大洋彼岸的一笔交易,让这个问题有了具体的价格。Google出价1000万美元,竞购停飞航空公司Spirit留下的部分内部数据,包括邮件、聊天、文档及部分软件。航空公司没能继续经营,员工多年处理订单、协调业务、解决问题留下的记录,却有了新的买家。