Anthropic Claude Fable 5 Now Available on AWS: Mythos-Class Performance with Built-in Safeguards

Anthropic has released Claude Fable 5 for use on Amazon Bedrock and Claude Platform on AWS. This model achieves state-of-the-art performance on nearly all benchmarks, excelling in software engineering, knowledge work, and vision tasks.

Key features of Claude Fable 5 include:

• Long-running asynchronous execution: Can sustain complex tasks such as coding and knowledge work for extended periods without intervention.
• Advanced vision capabilities: Understands diagrams, charts, and tables within files and PDFs, implements designs with high precision, and self-reviews its outputs.
• Self-verification capabilities: Updates its own skills based on what it learns and builds its own evaluation environments.

For safety, prompts with high risk of misuse—such as those related to cybersecurity, biology, chemistry, and health—are automatically routed to Opus 4.8 for response. Claude Mythos 5, which offers unrestricted performance, will be available only to a small number of verified customers.

Currently, Amazon Bedrock supports only programmatic access via the Anthropic Messages API, and the data sharing option must be enabled through the Data Retention API in order to use the model.