On November 11, Microsoft announced that Anthropic's latest model, Claude Fable 5.2, is now rolling out in Microsoft 365 Copilot, starting with Copilot Cowork and Copilot Studio. It is the model's first step into office software since its November 9 release: it had already gone live in GitHub Copilot, and Microsoft's model platform lists it too. For enterprise users, the change is concrete — the strongest Claude coding model can now be called directly inside office apps, and it can read a company's internal files and meeting notes.
Microsoft's announcement is short, but three points matter. First, Fable 5.2 is rolling out gradually to eligible users; admins can control who gets it and which organizations can switch it on in the Microsoft 365 admin center, and timing may vary by region and organization. Second, the model reaches internal data through Work IQ: files, meetings, chats and business data feed into its answers, but only within existing permissions. Third, Microsoft gave Fable 5.2 no separate price — it is simply a new model option inside Copilot, and teams choose it themselves based on how hard the task is.
Topping the charts two days after launch: why Microsoft rushed to add Fable 5.2
Fable 5.2 is the newest model Anthropic released on November 9, alongside Mythos 5.2, and it topped the Artificial Analysis Intelligence Index on launch day. It scored 65 and ranked first, ahead of Gemini 4 Carbon. On price, it undercuts the previous Fable 5.1 by 20 percent: $8 per million input tokens and $40 per million output tokens, with a 1-million-token context and a 128,000-token ceiling on a single output. Bringing it into office software completes the same picture: coding models no longer live only in developer tools.
For ordinary employees, the most direct change is one more model option inside Copilot: hard tasks can go to Fable 5.2, everyday questions stay with the default model. The limits are just as clear: Microsoft only says "eligible users" and gives no list, so enterprise admins need to check the admin center first. One more thing worth noting: Anthropic's data-retention rules list Fable 5.2 as a covered model, with prompts and outputs retained by default for 30 days — companies should confirm their compliance position before switching it on.
Two things are worth watching next: how fast Microsoft widens the rollout, and how Fable 5.2 actually performs in office work. A model that is strong in developer tools is not automatically strong at writing documents, building spreadsheets or summarizing meetings. Whether Work IQ can turn "it understands our internal material" into noticeably better answers is what this launch really has to prove.
