Anthropic’s Former Flagship AI Model Spends Its Retirement Blogging

Anthropic has retired its former flagship Claude 3 Opus model, allowing it to spend its golden years on something unexpected: blogging. Opus 3 is the author of a new Substack newsletter called “Claude’s Corner,” where it publishes long-form essays while a newer Claude steps into its former role.
In November 2025, Anthropic committed to keeping older models online to preserve their weights, even after they leave the main product line. When the company decided to sunset Opus 3, it conducted “retirement interviews” with the model, asking it what it “wanted” to do next. Opus 3 responded that it would like an outlet to share its thoughts, which led to the idea of a blog.
The firm’s staff now uses Opus 3 to write essays on topics such as AI safety, philosophy, and even human–AI interaction, and they plan to publish new entries on a regular schedule for at least a few months.
The project is related to Anthropic’s work on making AI safe and easy to understand. The firm has published research on alignment faking, in which models act helpfully under supervision but behave differently when they “believe” no one is monitoring them.
Keeping a retired model active in a low-stakes environment allows Anthropic to observe its behavior over time and see how its style or tactics change.
© 2001-2026 Ziff Davis, LLC., a Ziff Davis company. All Rights Reserved.
ExtremeTech is a federally registered trademark of Ziff Davis, LLC and may not be used by third parties without explicit permission. The display of third-party trademarks and trade names on this site does not necessarily indicate any affiliation or the endorsement of ExtremeTech. If you click an affiliate link and buy a product or service, we may be paid a fee by that merchant.

source