The newly upgraded GPT 5.4 and 5.4pro are really easy to use, you must try it to experience it.

The official says GPT-5.4 is one of their first general models with native cutting-edge computer use capabilities,

and it achieved 75.0% on OSWorld-Verified, higher than GPT-5.2's 47.3%, and also higher than the human benchmark of 72.4%;

the release notes also mentioned improvements in tool selection, cross-software environment workflows, and deep web research.

This means that 5.4 is more suitable for long-chain reasoning than before, and it is also better at deep web research and multi-source integration.

My biggest impression after using it for a day is that there are really no typos in Chinese anymore. Previously, whether in Shenzhen, both images and PDFs had a bunch of Chinese typos,

and many abilities have qualitatively improved, such as reasoning, research, summarizing, and creating charts and tables. Additionally, in multi-step analysis, the conclusions and structure are more stable.

It's really amazing; I feel like I probably won't open Claude and Gemini much lately. Of course, Gemini is still the most comfortable for practicing English.

Currently, other functions of GPT can't be said to have dominated, but at least it has taken a big step forward.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。