EXPLORINEER
en

Archive — US — techcrunch

Anthropic’s Claude Fable 5 is a version of Mythos the public can access today

Anthropic has launched Claude Fable 5, the first publicly accessible version of its powerful Mythos model. While Fable 5 excels in software engineering and knowledge work, it has safety constraints in high-risk areas. The model will be available through a tiered subscription, with a focus on safety and a data retention policy. Anthropic aims to balance advanced AI development with keeping systems secure, amid concerns over AI's rapid self-improvement.

Key points

  • Launch of Claude Fable 5 enables public access to Mythos technology initially limited to partners.
  • Fable 5 includes safety limits and defers to Claude Opus 4.8 in high-risk areas like cybersecurity and biology.
  • Anthropic expanded access to hundreds of organizations, focusing on those managing critical infrastructure.
  • Fable 5 is undergoing tiered subscription rollout with free access until June 22, after which usage credit will be required.
  • The model has undergone extensive testing with no successful jailbreaks detected.
  • New data retention policy for traffic is introduced to prevent misuse and enhance security.
  • Pricing set at $10 per million input tokens and $50 per million output tokens, which may deter widespread use due to higher costs.
  • Early data shows that 95% of Fable sessions operate without deferring to Opus 4.8, indicating high reliability.

Why it matters

  • The launch reflects a trend in AI towards more public, responsible access to powerful models amid safety concerns.
  • In 2023, many AI labs are addressing safety and regulatory frameworks as AI systems advance.
  • Anthropic's call for coordinated AI safety measures relates to ongoing debates about the risk of recursive self-improvement in AI.
  • This launch is part of a broader trend in 2025 where enterprise costs of AI usage are scrutinized as organizations adapt to new technologies.

Claude Fable 5's launch marks a significant step in democratizing access to advanced AI while ensuring safety. With robust testing and a focus on responsible deployment, it sets a precedent for future AI models, balancing innovation with necessary precautions to mitigate risks.