Microsoft has recently announced that its Copilot AI application has completed a major upgrade for web, Windows 11, and mobile users, and the latest GPT5.2 model has been fully rolled out. This new version is named the "Smart Plus" mode, aiming to provide users with "expert-level" office support.
Compared to GPT5.1 (Smart Mode) launched last month, the core advantage of GPT5.2 lies in its powerful "Thinking" variant model. Microsoft emphasizes that this model performs exceptionally well when handling complex and lengthy office tasks. Whether it's building complex Excel spreadsheets, creating elegant PowerPoint presentations, or conducting in-depth code reviews and long-document summaries, GPT5.2 demonstrates a high level of professionalism.
According to the latest knowledge-based job task evaluation (GDPval) data, GPT5.2 has achieved performance that can match or even exceed human industry experts in more than 70% of test cases, a ratio far higher than the previous GPT-5 model's approximately 38.8%. In more rigorous technical benchmark tests, the model also set new records: it scored 100% in the AIME2025 Mathematics Invitational Competition, and significantly outperformed its predecessor in the software engineering test (SWE-Bench Pro).
