Microsoft adds Anthropic models to Copilot

Overview

Microsoft is expanding its Microsoft 365 Copilot beyond OpenAI, adding Anthropic’s Claude models and giving enterprise users more flexibility in how they deploy generative AI.

The update, rolled out Sept. 24, brings Anthropic’s Claude Sonnet 4 and Claude Opus 4.1 into Copilot for the first time. It marks a notable shift for Microsoft, a long‑time, multibillion‑dollar backer of OpenAI, as Copilot will now run on multiple large language model (LLM) families rather than relying solely on OpenAI technology.

Why does it matter?

Analysts say the move reflects what customers want: choice. R “Ray” Wang of Constellation Research said enterprises are increasingly standardizing on more than one model to match different use cases, and that Microsoft has been steadily preparing for a diversified approach.

Industry watchers view the multi‑model turn as an enterprise‑first strategy that deepens Microsoft’s hold on workplace AI. While ChatGPT remains the most widely used generative AI chatbot, the majority of its roughly 700 million users are on the free tier, and OpenAI’s enterprise traction appears uneven by comparison, analysts noted.

A shifting AI alliance landscape

The change lands amid a broader reshuffling of AI alliances. Earlier this week, Nvidia committed $100 billion to OpenAI in the form of chip supplies, while OpenAI and Oracle announced a five‑year, $300 billion agreement for Oracle to provide the compute capacity OpenAI needs to build and operate new models. Together, the developments suggest Microsoft and OpenAI are each broadening their partnerships.

Where does Claude fit inside the Copilot?

Microsoft emphasized that OpenAI models remain central to Copilot. The addition of Claude expands options in specific areas: customers can choose Claude Sonnet 4 and Opus 4.1 inside Microsoft’s Researcher reasoning agent and within Copilot Studio, the company’s platform for building and managing enterprise agents.

Extending the model garden to Copilot

The move also aligns with Microsoft’s long-running “model garden” approach in Azure AI, where customers can access multiple third-party and open-source models. Google offers a similar setup in Vertex AI, and AWS does so with Amazon Bedrock. What’s different now, analysts say, is that Microsoft is extending that choice directly into Copilot—an easy‑to‑use system embedded across Microsoft 365, one of the world’s most widely deployed enterprise software suites.

Open questions and optics

There are questions, however, about the scope of Microsoft’s arrangement with Anthropic. Anshel Sag of Moor Insights & Strategy called the addition logical, especially given Anthropic’s emphasis on AI safety and responsible AI, but said the scale of the partnership hasn’t been made public. Anthropic has raised significant capital and is valued at around $183 billion, still well below OpenAI’s estimated $500 billion valuation.

The bottom line

By making Copilot multi‑model and adding Anthropic’s Claude, Microsoft is positioning itself to meet varied enterprise needs while signalling that the generative AI ecosystem is entering a new phase—one defined less by exclusivity and more by mix‑and‑match flexibility.