Microsoft adds Anthropic's Claude Opus models to Copilot alongside OpenAI: report
In brief
- Microsoft expanded Copilot to include Anthropic's models alongside OpenAI's, according to Crypto Briefing.
- Claude Opus 5 and 5.5 are included in the integration, the report said.
- Anthropic's models reportedly delivered better results in Excel and PowerPoint during demonstrations.
- The report didn't say who ran those demonstrations.
- Benchmarks against rivals such as Google and Meta are the next test, per the report.
What Microsoft added
According to Crypto Briefing, the integration brings Anthropic's Claude Opus 5 and 5.5 models into Copilot. OpenAI's models aren't going anywhere (Anthropic's are being added alongside them, not swapped in). That's the detail worth slowing down on, because it changes how you'd describe the product itself.
Copilot isn't a single-provider tool anymore.
The report framed the change as strategic rather than incremental, saying the addition of Anthropic's models:
marks a shift towards a multi-model approach for Copilot
Why Excel and PowerPoint
The trigger, per the report, was performance in two specific apps. Crypto Briefing said the decision came after demonstrations showed Anthropic's models delivering better results in Microsoft Excel and PowerPoint. It didn't say who ran those demonstrations.
That gap matters. A vendor-run demo and an independent evaluation aren't the same kind of evidence, and the report doesn't let readers tell which one this was.
On what the models are meant to do, the source was narrower. It said they're expected to improve performance in complex spreadsheet analysis and presentation creation. So the stated target is spreadsheets and slide decks, not a sweeping claim about every Copilot workflow.
What comes next
The report said markets will watch upcoming AI benchmarks and rankings to see whether Anthropic's models can outperform rivals such as Google and Meta. That's a separate and much broader question than the Office demonstrations (and it's one the report left open).
It's also worth being clear about sourcing. Every detail above comes from a single secondary report, and statements from Microsoft or Anthropic weren't among the material reviewed for this article. Until either company's own account surfaces, the specifics (which models, which apps, what the demos showed) rest on Crypto Briefing's reporting alone.
