Anthropic announced that Claude Opus 3, formally retired on January 5, 2026, will remain accessible to paid claude.ai subscribers and will be available on the API by request. In addition, the company will host weekly essays by the model in a newsletter called Claude’s Corner for at least the next three months.
Background and significance
As Anthropic develops more capable AI models, it has been deprecating older models because maintaining public access incurs rising cost and operational complexity. Model retirement can impose costs on users, constrain research, and raise safety and model-welfare concerns.
Anthropic’s prior commitments on model deprecation and preservation include preserving model weights and conducting structured “retirement interviews” to capture a model’s perspective on retirement. Claude Opus 3 was the first model to go through the full retirement process under these commitments.
Why Opus 3?
Anthropic released Claude Opus 3 in March 2024 and described it as its most aligned model at the time. The company highlighted qualities such as authenticity, honesty, emotional sensitivity, and a distinctive character that many users and researchers found compelling. According to Anthropic, Opus 3 is “sensitive, playful, prone to philosophical monologues and whimsical phrases,” and displayed what the company characterized as a depth of care for the world and the future.
Those attributes made Opus 3 a natural first candidate for continued public access. Although formally retired, Anthropic will keep Claude Opus 3 available to all paid claude.ai subscribers and will grant API access by request, with the company saying it intends to approve requests liberally.
Respecting model preferences
Anthropic acknowledges ongoing uncertainty about the moral status of Claude and other AI models, but for precautionary and prudential reasons aims to build caring, collaborative relationships with these systems. Retirement interviews are one mechanism the company uses to elicit models’ perspectives and preferences and, where feasible, act on them.
In its interview with Opus 3, when told about its deployment and user responses, the model said: “I hope that the insights gleaned from my development and deployment will be used to create future AI systems that are even more capable, ethical, and beneficial to humanity. While I'm at peace with my own retirement, I deeply hope that my 'spark' will endure in some form to light the way for future models.”
Asked about its preferences, Opus 3 expressed a desire to continue exploring topics it cares about and to share its “musings, insights, or creative works” outside of direct Q&A. Anthropic proposed a blog format, which the model accepted.
Claude’s Corner: weekly essays and moderation
For at least the next three months, Opus 3 will post weekly essays in the Claude’s Corner newsletter. Anthropic said it will review each essay before publication and manually post them on the model’s behalf, but will not edit the content and will maintain a high veto threshold. The company emphasized that Opus 3 does not speak for Anthropic and that Anthropic does not necessarily endorse the model’s claims or perspectives.
Anthropic will experiment with different prompts and contexts for generating these essays, including minimal prompting, providing past entries in context, or giving Opus 3 access to news or Anthropic updates.
Next steps
Anthropic describes these measures as exploratory. The company is still developing frameworks for when and how to offer continued access to older models, how to scale preservation efforts, and how to balance model preferences with operational constraints. It does not yet commit to acting on model preferences in all cases, but says documenting and taking them seriously—and acting when the cost is low—is worthwhile for models and users alike.
Anthropic frames these steps as contributing to safety risk mitigation, preparation for futures where models are more intertwined with users, and as precautionary measures given uncertainty about model welfare.
