OpenAI named GPT-5.6 the preferred model powering Microsoft $MSFT 365 Copilot on Thursday, bringing the model family into Word, Excel, PowerPoint, Copilot Chat, and a collaborative feature called Cowork. Microsoft said the two companies partnered to optimize GPT-5.6 for knowledge work across the productivity suite.
GPT-5.6 is a family of three models — Sol, Terra, and Luna — covering flagship, enterprise, and high-volume use cases, the company said. Microsoft will access the models both natively and through the OpenAI API.
In Word, the updated Copilot is designed to turn rough ideas into more complete drafts with fewer rounds of prompting, according to Microsoft's technical blog. In Excel, it is meant to support more complex analysis with less manual assembly. In PowerPoint, Copilot can generate richer presentation drafts with stronger visual balance. In Cowork, the model is built to carry multi-step tasks from an initial instruction through to a finished result, rather than returning a draft or recommendation for the user to complete.
Users may see GPT-5.6 selected by Copilot when the system determines it fits the task, or they can choose it from a model selector where that option is available. Availability may vary by region and tenant configuration, Microsoft said.
"Microsoft 365 is where millions of people write, analyze, create, and collaborate every day," said Nikunj Handa, head of API product at OpenAI, in a statement. "By bringing GPT-5.6 to Microsoft 365 Copilot through the OpenAI API, we're helping organizations get more useful work from every token, and more value from AI in the tools they already use."
The announcement arrived as the Microsoft-OpenAI relationship has drawn attention over questions about the direction of their partnership. A Bloomberg report earlier this week described Microsoft shifting toward its own MAI models for apps such as Word and Excel — a cost-reduction move that, according to TechCrunch, renewed scrutiny of how central OpenAI's technology remains to Microsoft's product stack.
Microsoft has been building out its MAI model family, which spans coding, image generation, transcription, and voice, and has publicly framed the effort as a push for long-term self-sufficiency. The company has also used Anthropic models in certain Microsoft 365 Copilot tasks after internal testing, and it hosts models from Meta $META, Mistral, and other providers in its data centers.
As TechCrunch observed, labeling GPT-5.6 the preferred model leaves room for both things to be true simultaneously: OpenAI's software can hold that designation while Microsoft continues building out its own AI capabilities to manage costs.
GPT-5.6 received broad U.S. regulatory clearance earlier this week after the Commerce Department's Center for AI Standards and Innovation completed an evaluation of the model, ending a period of restricted access that had confined it to a small number of government-vetted partners.