Anthropic has released Claude Fable 5.1 and Claude Mythos 5.1 as its new flagship models for coding and knowledge work, positioning them as capable of autonomous, multi-step tasks. The release includes a significant reduction in cache read pricing to $0.25 per million tokens, alongside a 1 million token context window.
- Fable 5.1 achieves an Artificial Analysis Intelligence Index score of 66 and 59.1% on HLE, outperforming previous versions.
- Cache read prices were cut by 75%, though the model uses approximately 1.7 times more output tokens than Fable 5.
- Community analysis suggests Fable and Mythos 5.1 may share underlying weights with different safety routing behaviors.
- Third-party platforms including Perplexity, OpenRouter, and Nous Portal have added support for the new models.
The update aims to address prior criticisms of impracticality by improving tone and verbosity while introducing Enterprise Frontier Safeguards for better agent observability.