OpenAI says GPT 5.6 is the ‘preferred model’ for Microsoft Copilot 365 amid breakup chatter

OpenAI has moved to quash speculation that its partnership with Microsoft is unraveling, declaring that its newest model line will remain central to Microsoft's flagship workplace software. The announcement arrived alongside the debut of GPT 5.6, which OpenAI billed as the version its systems will now favor when running inside Microsoft 365 Copilot.

The timing was hardly coincidental. Just days earlier, Bloomberg reported that Microsoft was quietly swapping out some OpenAI-built components across its productivity lineup in favor of homegrown alternatives, a move framed as a cost-saving measure. According to that report, Microsoft's own models — developed under the MAI label — have been steadily taking over duties in tools such as Word and Excel, chipping away at OpenAI's footprint in software used by hundreds of millions of people daily.

That reporting reignited a question that has followed the two companies for months: whether a relationship once viewed as a tightly bound alliance is starting to loosen. Microsoft and OpenAI built their partnership on billions of dollars in investment and deep technical integration, but the companies have lately sent conflicting signals about how entangled they still intend to be, fueling persistent chatter about a slow-motion separation.

OpenAI's messaging this week appears designed to counter that narrative directly. By positioning GPT 5.6 as the go-to engine behind Copilot for Microsoft 365, the company is signaling that, at least for now, its technology continues to occupy a privileged place inside Microsoft's commercial software ecosystem rather than being phased out in favor of in-house options.

The stakes extend well beyond a single product announcement. Microsoft 365 Copilot is embedded in the productivity suite that anchors much of corporate computing worldwide, meaning decisions about which underlying model powers it carry significant weight for both companies' business strategies. For OpenAI, retaining a foothold there helps preserve a major distribution channel and revenue stream. For Microsoft, the calculus increasingly involves balancing its reliance on an outside partner against the appeal of cheaper, self-built alternatives it can control end to end.

That tension helps explain why the two accounts — Microsoft trimming OpenAI's role in some products while OpenAI touts a "preferred" status in others — are not necessarily contradictory. Large technology companies frequently run parallel model strategies, testing internally developed systems in some features while continuing to lean on outside partners elsewhere, particularly for capabilities considered cutting edge. Microsoft's expanding MAI effort and its continued promotion of OpenAI-powered Copilot features could simply reflect a diversification strategy rather than a clean break.

Even so, the juxtaposition of the two reports underscores how closely industry observers are watching for cracks in a partnership that has helped define the generative AI boom. Microsoft has invested heavily in OpenAI and built much of its AI product roadmap around access to its models, while OpenAI has depended on Microsoft's cloud infrastructure and distribution reach. Any sign that either side is hedging tends to draw outsized attention given how much of the broader AI market's trajectory has been shaped by their collaboration.

Neither company has publicly detailed a roadmap for how model sourcing across Microsoft's products might evolve going forward, leaving open the possibility that the balance between OpenAI's models and Microsoft's in-house alternatives will continue to shift product by product. For now, GPT 5.6's designation as Copilot's favored model stands as OpenAI's clearest public signal that it does not view its role in Microsoft's software as diminishing, even as questions about the partnership's long-term shape persist.

Source: TechCrunch