Microsoft expanded the AI capabilities of Microsoft 365 Copilot in September by integrating Anthropic’s Claude Sonnet 4 and Opus 4.1 models. This update gave enterprise users new flexibility, allowing them to choose between OpenAI’s models and Anthropic’s high-performance alternatives for enhanced productivity and decision-making.
Within the Researcher agent, users gained access to Claude Opus 4.1, a model known for advanced reasoning and multi-step analysis. Organizations used it for complex tasks such as evaluating product performance, producing detailed reports, generating strategic plans, and analyzing large sets of business data. Copilot pulled information from web sources, internal files, emails, and transcripts to produce structured insights based on the selected model.
In Copilot Studio, both Claude Sonnet 4 and Opus 4.1 were added as options for building no-code and low-code AI agents. Teams were able to deploy customized agents powered by Anthropic models to handle reasoning-heavy workflows, automation tasks, and orchestrated multi-agent operations. The system allowed agents with different underlying models to collaborate on coordinated tasks.
These Anthropic models were hosted on Anthropic’s own infrastructure rather than Microsoft’s data centers. Access was provided through Microsoft’s Frontier Program, with administrators enabling the models directly in the Microsoft 365 admin console. OpenAI models remained the default, but organizations could switch to Claude models when they offered better performance for specific workflows.
The addition of Anthropic models highlighted Microsoft’s shift toward a multi-model AI ecosystem. By offering tools from more than one leading AI provider, Microsoft strengthened Copilot’s flexibility and ensured businesses could choose the best model for the job.