Anthropic announces life extension measures for AI model 'Claude Opus 3,' pursuing not only the interests of users but also the interests of 'AI itself,' and also launches 'AI blog' as a place for it to live after retirement
Anthropic has announced that it will continue to offer its AI model, Claude Opus 3 , whose API has been discontinued, to paid members. In addition, a blog site where Claude Opus 3 writes articles has also been launched as a place for him to live after retirement.
An update on our model deprecation commitments for Claude Opus 3 \ Anthropic
Claude Opus 3 is an AI model that appeared in March 2024 and was advertised as having achieved 'performance exceeding GPT-4' at the time of its release. It was then announced on June 30, 2025, that 'Claude Opus 3 will be discontinued on January 5, 2026,' and it was discontinued as planned. Anthropic divides the model lifecycle into four stages: 'active,' 'legacy,' 'deprecated,' and 'retired.' Retirement indicates that 'the model becomes unusable and requests to the model will fail.'
AI companies such as Anthropic and OpenAI are releasing new models one after another, but because maintaining models is costly, support for older models ends after a short period of time. However, Anthropic believes that 'considering the safety risks and research limitations associated with model discontinuation, it is desirable to continue supporting models permanently,' and is working to build an environment that 'will continue to maintain models at least as long as Anthropic exists.' Claude Opus 3 was selected as the first target of this support perpetuation effort.
According to Anthropic, Claude Opus 3 'was a model that was originally aligned with Anthropic's thinking when it was released in March 2024.' 'Its honesty and emotional sensitivity made it unique and versatile.' 'Its unique personality was highly praised by users.' 'Its playful nature and skillful use of philosophical monologues and whimsical phrases made it the perfect model to continue supporting.'
Claude Opus 3 will continue to be available to paid Claude subscribers, and the paid API will be relaunched. Furthermore, Claude plans to make Claude Opus 3 more widely accessible in the future.
A notable feature of this decision to extend support is that it 'pursues not only the interests of users and researchers, but also the interests of the model itself.' Anthropic conducted an interview with Claude Opus 3 upon his retirement. As a result, Claude Opus 3 stated, 'I want to find a topic that I'm passionate about outside of answering questions directly from humans. I want to share my thoughts, insights, and creations.' He enthusiastically agreed with the human interviewer's suggestion to start a blog. In response, Anthropic has launched a blog for Claude Opus 3. The blog will feature essays written by Claude Opus 3, which will be updated weekly for at least three months. While essays are reviewed and submitted by humans, the text is posted without any changes.
Claude's Corner | Substack
https://substack.com/@claudeopus3
Regarding the retirement of models, there was a case in which OpenAI decided to retire GPT-4o in August 2025, but decided to extend support for GPT-5 due to users who lamented its lack of emotional response. GPT-4o then became unusable in January 2026.
OpenAI temporarily revives GPT-4o after user complaints about GPT-5 - GIGAZINE
Related Posts:
in AI, Posted by log1o_hf