Microsoft MS365 Copilot gets a new core: integrating Anthropic Claude models

Microsoft recently announced a major update to its Microsoft 365 Copilot The model library has officially ushered in the Anthropic company's series of models. This means that the long-standing modeling of the series, which has been done primarily by OpenAI model-driven CopilotThe company is now offering a second top model option for enterprise users. This initiative is not only a simple technology overlay, but also signifies that Microsoft's enterprise AI strategy is shifting from a single dependency to a more open and diversified "multi-model" route, aiming to provide the most appropriate tools for different business scenarios, and also intensifying the competition for the top AI models in the enterprise services market.

first batch of integrated Anthropic The application of the model will be Microsoft 365 Copilot hit the nail on the head Researcher (research assistants) and Copilot Studio (Intelligent Body Studio). Business users will be able to invoke both tools in the Claude Sonnet 4 cap (a poem) Claude Opus 4.1 Model.

Built for Deep Research: Claude Opus 4.1 in Researcher

Researcher act as Microsoft 365 Copilot The reasoning intelligences used to perform complex analytical tasks in the OpenAI model and Anthropic The flagship model of the Claude Opus 4.1 Choose between.

Claude Opus 4.1 constitute Anthropic The most capable model in the product line is particularly good at handling tasks that require deep reasoning, long contextual understanding, and complex instruction following. This makes it one of the Researcher the ideal partner. For example, when developing a quarterly market entry strategy, theResearcher This can be done in the Opus 4.1 The company is driven to simultaneously analyze hundreds of pages of industry reports, internal email communications, Teams meeting minutes, and sales data files, and extract coherent and deep insights from them. Model selection allows organizations to choose the most appropriate "brain" based on the complexity of the task and the reasoning power required.

Customizing Enterprise Intelligence: Claude Model Options in Copilot Studio

In low-code development platforms Copilot Studio in which developers now have the option of using the Claude Sonnet 4 maybe Claude Opus 4.1 to build and manage customized enterprise-grade AI intelligences.

together with Opus 4.1 The quest for extreme performance varies.Claude Sonnet 4 The model strikes a better balance between performance, speed and cost, making it ideal for automated processes that require rapid response and large-scale deployment. Developers can use the Copilot Studio utilization Sonnet 4 Build intelligences for tasks such as customer service, internal process automation, or data extraction to achieve high efficiency and cost control.

More valuable.Copilot Studio Support for multi-intelligence systems allows developers to mix and match models from different companies. For example, a complex business process can be designed such that the front-end customer interaction session uses a responsive Claude Sonnet 4The backend, which requires in-depth data analysis, calls the OpenAI (used form a nominal expression) GPT Model or Claude Opus 4.1. This flexibility allows organizations to match the best AI model for each specific task as if they were assembling a team of expert consultants.

The operation of switching models has been designed to be very convenient. In the Researcher in which the user can switch with a built-in switch; in the Copilot Studio In this case, you can select the desired model directly from the drop-down menu.

How do I get started?

According to Microsoft's announcement, the update will be available through its Frontier Program Progressively open to clients.

• Claude in Researcher: Ownership Microsoft 365 Copilot Licensed commercial customers who opt in to the Frontier Program After that, access will be granted in batches.
• Claude in Copilot Studio:: Again, the client needs to actively opt in to begin experimenting with a program based on Anthropic Model building intelligences.

Frontier Program is an early-experience program set up by Microsoft for enterprise customers to try out the latest AI technology and provide feedback before the feature is fully available. It's important to note that, according to the Microsoft noteAnthropic The model will be hosted outside of the Microsoft-managed environment and follow the Anthropic The terms of service of the

This feature is only available for Business Edition subscribers Microsoft 365 Copilot users of the personal or home version of the Copilot Pro Subscriptions are not supported. The feature is expected to complete its push to all eligible users within a month, and supports the world's major languages, including Simplified Chinese.