Microsoft Copilot Claude models

Microsoft has widened its AI options by adding Anthropic’s Claude models to Copilot, signalling a strategic shift away from sole reliance on OpenAI.

Microsoft Copilot now supports Anthropic’s Claude Sonnet 4 and Claude Opus 4.1 alongside OpenAI models,Researcher, a Copilot agent for complex tasks, lets subscribers “Try Claude” instead of defaulting to OpenAI,Copilot Studio customers can mix and match Claude with OpenAI and other models hosted on Azure

Microsoft opens the Claude door

Microsoft has announced that two of Anthropic’s frontier models, Claude Sonnet 4 and Claude Opus 4.1, are now available across its Copilot ecosystem. Until now, Copilot customers had largely been limited to OpenAI’s GPT series, with GPT-4o serving as the default.

Charles Lamanna, president of Copilot’s business and industry division, described the move as giving customers more choice, rather than a replacement of existing models. He noted that OpenAI’s latest models would remain central to Copilot, but businesses could now select Anthropic’s alternatives if they fit better with their needs.

Researcher gains Claude’s reasoning power

The most visible addition is Claude Opus 4.1 in Researcher, a Copilot agent designed for multi-step tasks. Within Microsoft 365, subscribers will see a new “Try Claude” button, allowing them to switch from OpenAI’s default to Anthropic’s model.

For knowledge workers, this means the ability to compare model performance on tasks such as drafting proposals, synthesising research, or generating structured reports. The appeal lies in the fact that Opus 4.1 is regarded as one of the strongest models for extended reasoning; a capability much in demand among analysts, consultants, and strategists.

Studio flexibility widens the field

Claude Sonnet 4 and Opus 4.1 are also now integrated into Copilot Studio, Microsoft’s custom agent-building platform. Studio users can already assemble agents using OpenAI’s models, but the update introduces a new dropdown option under “Agent’s model,” giving direct access to the Claude family.

Importantly, Copilot Studio allows hybrid agent designs. This means a single workflow could, for example, use OpenAI for summarisation while turning to Anthropic’s models for structured reasoning. Both Claude models, however, are categorised as “external” since they are not hosted by Microsoft’s servers and remain under Anthropic’s terms of service.

A careful step back from OpenAI

The partnership reflects shifting dynamics in the AI sector. For much of the past two years, Microsoft’s fortunes in AI were bound tightly to OpenAI, including a multi-billion dollar investment and priority access to its models. But the landscape is evolving. OpenAI is rumoured to be working more closely with Google Cloud, while Microsoft has launched its own experimental MAI models and formed fresh partnerships.

Anthropic, meanwhile, has emerged as a credible rival in enterprise AI, with a valuation topping $183 billion after its most recent funding round. Its products have gained a reputation for reliability and safe deployment at scale, factors likely to appeal to Microsoft’s corporate clientele. This development mirrors the growing trend of executives treading carefully on generative AI adoption, seeking robust and trustworthy solutions.

What this means for Asia-Pacific users

For businesses in Singapore, Tokyo, Sydney or Jakarta, the shift is about choice and flexibility. Many regional enterprises face complex compliance and linguistic challenges, from multilingual customer service to regulatory filings. Having both OpenAI and Anthropic in the same platform allows teams to benchmark performance in their local context before making strategic commitments. This aligns with the broader trend of APAC AI in 2026: 4 Trends You Need To Know, where regional nuances are becoming increasingly important.

The question, of course, is whether this widening of options will increase costs or complexity. At present, Microsoft appears keen to frame it as customer empowerment rather than a shake-up of pricing. But as Anthropic’s models are listed as external, firms may need to adapt to an additional set of usage terms. For insight into how other regions are approaching AI governance, consider Taiwan’s AI Law Is Quietly Redefining What “Responsible Innovation” Means.

The addition of Claude models to Copilot is more than a simple feature update. It signals Microsoft’s intention to decouple its AI ambitions from a single partner, while giving enterprises a broader palette of tools to experiment with. For professionals in Asia and beyond, it sets the stage for a new era where choice and flexibility will be as valuable as raw AI power. This move also highlights the increasing competition in the AI space, as detailed in reports like the Stanford AI Index, which tracks the global AI landscape and investment trends.

What do you think? Will companies actually use both OpenAI and Anthropic side by side, or will loyalties quickly settle around one?