OpenAI Says GPT 5.6 Will Remain Microsoft 365 Copilot's "Preferred Model"

OpenAI is moving to clear up questions about its relationship with Microsoft. During the launch of GPT 5.6 on Thursday, the company announced that the model would become the "preferred model" powering Microsoft 365 Copilot, the AI layer built into Microsoft's productivity apps. The announcement lands at a moment when the partnership between the two companies had seemed to be cooling.

A denial with caveats

The timing of the announcement is no coincidence. Earlier this week, Bloomberg reported that Microsoft was replacing some of OpenAI's software with its own in-house models to cut costs. Those in-house models, known as MAI, were reportedly being used to power apps like Word and Excel.

The report fueled a question that keeps coming up about the two companies: are they drifting apart? For a while, OpenAI and Microsoft were practically inseparable, but lately they've been sending mixed signals about the state of their relationship.

OpenAI's response was to position GPT 5.6 as the go-to model for Copilot. In a related blog post, the company said GPT 5.6 would support Microsoft users across the company's suite of productivity apps, including Word, Excel, PowerPoint, and Cowork.

"Our partnership with Microsoft has always been about bringing the benefits of advanced AI to more individuals and organizations, and we're excited to continue building on that shared commitment," OpenAI wrote.

What "preferred model" actually means

Here's the problem with the announcement: it's not entirely clear what being a "preferred model" entails, beyond the fact that OpenAI's software will keep powering Microsoft's apps. It's a label that sounds like a firm commitment, but its actual scope is left undefined.

The fine print matters here. It was never reported that ChatGPT's software would stop powering Microsoft's apps — only that Microsoft was leaning more heavily on its own technology to cut costs. The new "preferred model" disclosure doesn't contradict that earlier reporting; both things can be true at once.

In other words, GPT 5.6 being the "preferred" model doesn't rule out Microsoft continuing to use its own MAI models in parallel. A company can have a preferred vendor while quietly building internal alternatives at the same time. In fact, that's precisely the logic behind reducing dependence on a partner without breaking away all at once.

Why this relationship matters

The OpenAI-Microsoft alliance has been one of the pillars of the generative AI market. Microsoft has woven OpenAI's technology into much of its product lineup, while OpenAI has relied on Microsoft's infrastructure and backing to scale. Any hint of distance between the two, however small, gets scrutinized closely because it affects how power is distributed across the industry.

Microsoft's push toward its own models reflects a very concrete, very familiar motivation: cost. Running AI models at scale is expensive, and depending on a third party for features that touch millions of Word and Excel users adds up to a hefty bill. Building in-house technology is a way to control that spending and avoid being locked into a single vendor's terms.

For OpenAI, keeping Microsoft as both a customer and a showcase is strategic. Losing ground in the world's most widely used corporate productivity ecosystem would be a reputational and commercial blow. That's likely why the GPT 5.6 "preferred model" announcement came right on the heels of Bloomberg's report — it's a move designed to reassure the market.

A backdrop of broader tension

The announcement doesn't exist in a vacuum. In recent days, OpenAI has made headlines on several fronts that paint a delicate picture for the company, from shakeups in its executive ranks to legal disputes with other tech giants. A product launch paired with a reassuring message about Microsoft fits neatly into that broader pattern of shifting alliances.

What's emerging is a picture of two partners that are no longer joined at the hip. Microsoft wants technological independence and cost control; OpenAI wants to hold on to its privileged position within the Redmond giant's ecosystem. Those goals are compatible up to a point, and the balance between them is what defines the relationship today.

What to watch next

The open question is how much of Microsoft's app traffic will keep running on GPT 5.6 versus how much will shift to MAI models. The "preferred" label doesn't answer that. It remains to be seen whether Microsoft will eventually spell out which specific features rely on which model, and in what proportion.

For everyday users of Word, Excel, PowerPoint, or Cowork, the change will likely be invisible — the AI assistance will still be there. What's at stake behind the scenes is who controls the technology making it work, and that division of power isn't settled yet. OpenAI's announcement may put the breakup headlines to rest, but not the underlying question.