Claude Fable 5.1 and Mythos 5.1 Overview

Model Performance and Efficiency

Claude Fable 5.1 offers a significant shift in cost and performance compared to its predecessor, Fable 5. While the base pricing remains $10 per million input tokens and $50 per million output tokens, the model achieves lower overall costs through increased efficiency in cache reads. For typical workflows, users can expect approximately 25% lower costs, with highly agentic tasks potentially seeing savings up to 50%. Benchmarks provided by the developer indicate that Fable 5.1 outperforms Fable 5 across multidisciplinary reasoning and agentic terminal coding tasks, even when comparing lower-tier Fable 5.1 configurations against high-tier Fable 5 settings.

Safeguards and Availability

Alongside the release of Fable 5.1, Anthropic introduced Mythos 5.1, a variant designed for vetted organizations in cybersecurity and life sciences. Mythos 5.1 features more permissive safeguards for specialized use cases and is accessible only through specific verification programs. Both models are available immediately via the desktop application, terminal, and cloud providers including Amazon Web Services, Google Cloud, and Microsoft Azure. Additionally, the update includes an EFS system for enterprise customers, allowing for an opt-in configuration that ensures no data retention.

Preliminary Testing

In a basic test generating a rotating 3D cartoon bear, Fable 5.1 produced a result with more realistic shadows and physics compared to Fable 5. A cost comparison on this specific task showed Fable 5.1 cost $4.53, while the Fable 5 version cost $5.41, confirming a marginal reduction in expense for the same prompt.