Microsoft adds auto AI in VS Code, favoring Anthropic over OpenAI

Microsoft is rolling out a new feature in Visual Studio Code that automatically selects the best AI model for developers using GitHub Copilot. The “auto model” function will dynamically pick between options like Claude Sonnet 4, GPT-5, GPT-5 Mini, and others to deliver what Microsoft calls “optimal performance.”

For free GitHub Copilot users, the model will shift depending on the task. However, paid subscribers will primarily rely on Anthropic’s Claude Sonnet 4—an explicit sign that Microsoft is leaning more heavily on Anthropic’s technology for coding tasks.

Microsoft Quietly Favors Claude Sonnet 4

According to sources familiar with the company’s developer strategy, Microsoft has been directing its engineers toward Claude Sonnet 4 in recent months. An internal email sent in June by Julia Liuson, head of Microsoft’s developer division, stated:

“Based on internal benchmarks, Claude Sonnet 4 is our recommended model for GitHub Copilot.”

Although the email came before the launch of GPT-5, insiders say Microsoft’s guidance has not shifted even after OpenAI’s latest release.

This preference underscores Microsoft’s growing reliance on Anthropic’s models, despite its massive financial and strategic investment in OpenAI.

Microsoft Building Its Own AI Capabilities

Beyond choosing between external partners, Microsoft is also ramping up efforts to train its own models. At a recent employee town hall, Mustafa Suleyman, Microsoft’s AI chief, highlighted the company’s internal ambitions:

“We’re also going to be making significant investments in our own cluster. So today, MAI-1-preview was only trained on 15,000 H100s, a tiny cluster in the grand scheme of things.”

This signals a future where Microsoft could reduce dependency on external AI vendors by scaling up proprietary models.

Anthropic in Microsoft 365

The shift toward Anthropic is not limited to developer tools. According to The Information, Microsoft plans to integrate Anthropic’s AI models into its Microsoft 365 suite, including Excel and PowerPoint. Tests reportedly showed Anthropic models outperforming OpenAI’s technology in certain productivity tasks.

If true, this would represent a significant diversification of Microsoft’s AI ecosystem—bringing in Anthropic alongside OpenAI for its flagship productivity apps.

The Complex Microsoft–OpenAI Partnership

The timing of these moves is noteworthy. Just last week, Microsoft and OpenAI announced a new agreement that could help pave the way for OpenAI’s initial public offering. Microsoft has poured more than $13 billion into OpenAI since 2019, but the relationship is complicated by revenue-sharing arrangements and competitive pressures.

Under the updated deal, OpenAI is allowed to use rival cloud providers in addition to Microsoft Azure. Meanwhile, Microsoft has promised more details on what it calls the “next phase” of its OpenAI relationship in the near future.

What It Means for Developers

For developers, the new auto model feature in Visual Studio Code means less decision-making over which model to use—Microsoft will handle that in the background. But the company’s heavy emphasis on Claude Sonnet 4 suggests that developers may see Anthropic’s fingerprints across GitHub Copilot and, increasingly, Microsoft 365.

This evolving strategy shows Microsoft is carefully balancing its dependence on OpenAI with both Anthropic’s strengths and its own AI ambitions—a three-pronged approach that could reshape its competitive position in the rapidly shifting AI landscape.