Microsoft Integrates GPT-5.2 into Copilot as Smart Plus Mode
Microsoft has revealed that its Copilot assistant will adopt OpenAI's new GPT-5.2 language model. This update, dubbed Smart Plus mode, will be rolled out in the web, Windows, and mobile versions. Users will be able to select between this new model and the current GPT-5.1 depending on the complexity of the task they want to perform. 🚀
A Leap in User Capabilities
The Smart Plus mode, powered by GPT-5.2, is not a simple incremental improvement. It introduces a set of features designed to handle more demanding tasks. The assistant can now generate complete office files, interact with programming code, and process visual and textual information more deeply.
Key actions enabled by GPT-5.2:- Generate spreadsheets and presentations from scratch, with structure and initial data.
- Write, debug, and verify code in multiple programming languages.
- Analyze extensive documents to extract specific data or synthesize their content.
- Connect with external tools and services to expand its functionality.
- Interpret images, describe their content, and generate text from them.
GPT-5.2 significantly improves performance in tasks that require complex reasoning and productivity workflows.
Stronger Reasoning and More Reliable Results
According to data shared by Microsoft, the core of GPT-5.2 shines in scenarios that demand multi-step reasoning and productivity-oriented workflows. This translates into responses with greater accuracy, superior ability to follow detailed instructions, and more consistent execution of long processes.
Areas where improvements are noticeable:- Accuracy and relevance in generated responses.
- Ability to adhere to complex and nuanced instructions.
- Reliability when handling tasks divided into multiple stages or decisions.
Model Selection and Coexistence
One of Microsoft's key strategies is to allow GPT-5.1 and GPT-5.2 to coexist within Copilot. This gives users the flexibility to choose the AI engine that best suits their immediate needs, whether for a quick query or an elaborate project. The company even suggests consulting the assistant itself to decide which model to use in specific situations, although its sense of humor for late excuses seems unchanged. 🤖