OpenAI used the launch of its new GPT-5.6 model family Thursday to make a specific claim about a specific partner: the model would become the “preferred model” powering Microsoft’s 365 Copilot across Word, Excel, PowerPoint, and Cowork. That announcement landed just days after reports surfaced that Microsoft was leaning more heavily on its own in-house models, known internally as MAI, to power some of those same apps in an effort to cut costs. A company naming itself “preferred” right after reports of the opposite trend is what NewsTrackerToday rests on as the actual story here, more than the GPT-5.6 launch itself.
The two companies have looked increasingly less inseparable in recent months, despite years of being treated as a package deal in the AI industry. Microsoft has been quietly diversifying its model lineup, and OpenAI has been building relationships with other major partners and infrastructure providers of its own. Neither company has stated outright that the relationship is ending, and neither has confirmed it’s staying exactly as it was either.
Sophie Leclerc, who covers the technology sector, unpacks what “preferred model” actually commits either company to: “It’s a marketing term, not a contractual one, at least based on what’s been disclosed publicly. OpenAI didn’t say GPT-5.6 would be the exclusive model, or the default model, or the only model Microsoft ships inside Copilot. It said ‘preferred.’ That word choice leaves enormous room for Microsoft to keep expanding its own MAI models in parallel without technically contradicting anything OpenAI announced this week.” That gap between a marketing word and a binding commitment is what NewsTrackerToday draws out as the detail this launch conveniently leaves undefined.
OpenAI’s own blog post framed the update as continuity rather than reassurance: “Our partnership with Microsoft has always been about bringing the benefits of advanced AI to more individuals and organizations,” the company wrote, language that reads as an extension of an existing relationship rather than a response to specific reporting about cost-cutting. Notably, nothing in that post directly addresses the MAI-model reporting from earlier in the week, or explains what changes for Microsoft users if the two claims are both true at once.
Isabella Moretti reads the corporate dynamics at play: “Two things can be true simultaneously here, and that’s exactly what makes this interesting. Microsoft can genuinely be building out MAI to reduce dependency and cost, and OpenAI’s GPT-5.6 can genuinely remain the flagship model inside Copilot for the highest-value tasks. Companies rarely make a clean binary break from a partner this deeply integrated into their product line. What’s more likely is a slow rebalancing where OpenAI keeps the premium positioning while Microsoft’s own models quietly absorb more of the routine workload.” That rebalancing, more than a breakup headline, is what NewsTrackerToday comes back to as the more accurate read of where the two companies actually stand.
It was never reported that ChatGPT’s underlying software would stop powering Microsoft’s productivity apps entirely, only that Microsoft was increasingly relying on its own models for some of that work. OpenAI’s “preferred model” disclosure this week doesn’t actually contradict that earlier reporting, it simply doesn’t confirm the scale of the shift either.
What actually changes for a Word or Excel user day to day, based on which model happens to be running behind the scenes at any given moment, remains entirely unclear from either company’s public statements this week. Whether OpenAI or Microsoft ever specifies what share of Copilot’s workload each company’s models are actually handling, rather than trading vague reassurance language back and forth, is what News Tracker Today leans on as the real disclosure investors and enterprise customers are still waiting on.
