Microsoft Copilot Now Runs ChatGPT and Claude at the Same Time

Microsoft just made its AI research tool significantly smarter. Instead of relying on a single AI model, Copilot’s Researcher tool can now tap into both OpenAI’s ChatGPT and Anthropic’s Claude simultaneously — and the results are genuinely impressive.

Think of it like getting a second opinion from a brilliant colleague. One AI drafts the answer, another checks it for gaps and errors, and you get something better than either could produce alone.

The Critique Feature Builds a Feedback Loop

Microsoft’s new Critique feature is the engine behind this upgrade. Here’s how it works: GPT generates an initial response, and then Claude reviews and refines it. Back and forth, round after round.

Microsoft calls it a “powerful feedback loop.” And honestly, that’s a fair description. The process mirrors how real academic and professional research works — first draft, peer review, revision.

The company says this architecture improves factual accuracy, analytical depth, and how well the final answer is presented. They also claim Researcher now scores higher than Perplexity’s Deep Research models on the Deep Research Accuracy, Completeness, and Objectivity benchmark. That’s a meaningful benchmark to beat, especially given how seriously Perplexity takes its research product.

Model Council Gives You Both Answers Side by Side

Critique isn’t the only new addition. Microsoft also introduced Model Council, which works quite differently.

Instead of combining the models behind the scenes, Model Council shows you responses from both ChatGPT and Claude side by side. You see where they agree. You see where they diverge. Then you decide what to make of it.

That transparency is genuinely useful. Sometimes the most interesting insight lives in the disagreement between two smart sources. So rather than hiding the process, Model Council puts it right in front of you.

If you prefer having more control over your research rather than letting AI synthesize everything automatically, Model Council is the better option. It hands you the raw material and trusts you to draw conclusions.

Where to Find These Features Right Now

Both Critique and Model Council are currently live inside Microsoft 365 Copilot’s Frontier program. Frontier is Microsoft’s early access space for new AI features — essentially a testing ground before broader rollout.

It’s worth noting that Anthropic already offers something similar on its own platform. Claude’s standalone Research feature uses multiple AI agents working together to handle complex questions. Microsoft is essentially building that same multi-agent philosophy directly into its productivity suite.

Why This Shift Matters for Everyday Research

The bigger story here isn’t just about two AI models talking to each other. It’s about where AI research tools are heading.

Single-model answers are fast, but they carry all the blind spots of that one model. Multi-model approaches catch more errors, cover more angles, and produce work that’s harder to poke holes in. That’s exactly what Critique and Model Council deliver.

For anyone who uses Microsoft 365 for serious research — consultants, analysts, writers, academics — this is a meaningful upgrade. The gap between “quick AI answer” and “actually reliable AI answer” just got a lot smaller.

If you have access to the Frontier program, both features are worth trying today. And if you don’t, they’re worth watching closely as Microsoft rolls them out more broadly.