Amid Talk of a Widening Rift, OpenAI Declares GPT-5.6 the New Face of Microsoft 365 Copilot

OpenAI introduced its newest model, GPT-5.6, on Thursday and simultaneously designated it as the "preferred model" running Microsoft 365 Copilot - a positioning statement that arrives just days after reports pointed to Microsoft scaling back its dependence on OpenAI systems.

According to the company, GPT-5.6 will now serve as the primary engine behind Copilot functionality in Word, Excel, PowerPoint, and Cowork. OpenAI framed the rollout in a blog post as a natural extension of its mission to make cutting-edge AI more accessible through Microsoft's product suite, and maintained that its collaboration with the software giant remains as strong as ever.

Where the speculation began

The timing traces back to a Bloomberg report published earlier in the week, which detailed how Microsoft had started weaving more of its proprietary AI models, internally referred to as MAI, into core Microsoft 365 tools such as Word and Excel. Per that reporting, the driving factor behind the shift was cost: running large-scale generative AI services is expensive, and Microsoft appeared to be looking for ways to trim that overhead by leaning on technology it built in-house.

This detail carries weight given the scale of Microsoft's financial commitment to OpenAI, which runs into billions of dollars, and the extent to which Copilot across Windows, Microsoft 365, and Azure has historically depended on OpenAI's models to function. Yet Microsoft's parallel investment in its own AI research division has been growing too, raising the possibility that the company is positioning itself to need OpenAI's technology less over time.

Together, these developments have led many in the tech world to reconsider whether the two companies' relationship is still best described as a partnership, or whether it's edging toward quiet competition.

Ambiguity remains over implementation

Despite the reassuring language, OpenAI's post left key questions unanswered. The company never clarified precisely what "preferred model" translates to operationally, nor did it lay out how work will actually be distributed between GPT-5.6 and Microsoft's own MAI-based systems moving forward.

It's also worth noting that Bloomberg's report never claimed Microsoft was severing its ties with OpenAI. Instead, it framed the shift as an expansion of Microsoft's internal model usage, driven primarily by efficiency and cost considerations rather than any move to phase OpenAI out entirely.

Taken together, the two reports aren't necessarily in conflict. If anything, OpenAI's statement complements rather than refutes Bloomberg's account confirming that GPT-5.6 will still occupy a central spot within Copilot, even as Microsoft continues to grow its own competing model lineup in the background.

Everyday users of Microsoft's productivity suite won't notice any disruption - GPT-5.6 will continue operating within the tools they already rely on daily. But from an industry standpoint, the episode illustrates a more nuanced trend: two companies can keep deepening their partnership on one front while simultaneously building independent capabilities on another, without that necessarily signaling a breakdown in the relationship. How this balance plays out in the months ahead may offer a clearer signal of where things are truly headed.

Read More: OpenAI Launches GPT-Live: New Voice Model Brings Real-Time Replies, Background Processing to ChatGPT