Anthropic Claude Fable 5 on AWS: Mythos-class capabilities with built-in safeguards now available

This article announces the availability of Claude Fable 5 on Amazon Bedrock and Claude Platform on AWS, featuring Mythos-level AI capabilities with built-in safeguards.

• Claude Fable 5 delivers state-of-the-art performance in software engineering, knowledge work, and vision tasks
• Supports long-running asynchronous execution for complex, sustained tasks without intervention
• Advanced vision capabilities understand diagrams, charts, tables in files and PDFs
• Includes proactive self-verification and skill development based on learnings
• Built-in safeguards limit performance on cybersecurity, biology, chemistry, health prompts
• Harmful prompts route to Opus 4.8 instead; you pay only Opus prices for routed requests
• Requires opt-in data sharing via Data Retention API; Anthropic requires 30-day retention
• Access via Anthropic Messages API, Invoke API, or Converse API on bedrock-runtime
• Available in US East (N. Virginia) and Europe (Stockholm) regions; gradual rollout to all AWS accounts

Claude Fable 5 brings powerful AI capabilities to AWS customers with safety guardrails, enabling ambitious long-running work across multiple domains.