Microsoft’s Copilot makes Anthropic’s Claude and OpenAI’s GPT team up

Microsoft added an optional critique feature to Copilot’s Researcher that chains two models for drafting and review.

With critique enabled, GPT writes the draft and Claude reviews for accuracy, completeness, and citation integrity.

Microsoft reported higher DRACO benchmark scores for Researcher with Critique than for individual models.

Microsoft also added a council view to compare model outputs and launched Copilot Cowork in the M365 Frontier early-access program.