Microsoft Brings Anthropic Into Copilot

Microsoft is integrating Anthropic's Claude models into Microsoft 365 Copilot, providing businesses with a multi-model AI choice alongside OpenAI. This new partnership introduces Claude Sonnet 4 and Claude Opus 4.1, which users can select on a per-task basis, with OpenAI remaining the default.

A new "Copilot Researcher" feature allows users to run the same prompt across different models and compare the results side-by-side within Microsoft 365 applications like Word and Excel. This is designed for testing and comparing outputs for tasks that require complex reasoning or careful summarization.

In Copilot Studio, administrators can now choose between Anthropic and OpenAI when creating automations and AI agents. IT teams have full control over these options through the Microsoft 365 admin center, allowing them to enable or block specific AI providers.

This strategic move by Microsoft formalizes a portfolio approach to AI. It aims to reduce reliance on a single vendor, accelerate feature development, and allow customers to choose the best model for their specific needs, from long-form analysis to cautious reasoning.

The immediate impact for users is greater flexibility in prompting. A knowledge worker can now easily switch between Claude and OpenAI for different documents or research tasks to see which model performs better.

For operations teams, this means they can route specific, reasoning-heavy automation steps to Claude while using OpenAI for other tasks. Admins now have granular control to manage and govern these AI providers at any time.

This move signals a strategic shift towards a "best tool wins" philosophy. While OpenAI remains central to Microsoft's AI strategy, Copilot is increasingly resembling a flexible marketplace for different AI models, with the company even willing to surface models hosted on AWS.