)
Anthropic launched Claude Fable 5.1 and Claude Mythos 5.1 on Tuesday, marking a major expansion of its most capable AI systems into general availability for corporate users and developers. According to official company disclosures, the two models share identical core architectures but employ vastly different safeguarding layers. While Fable 5.1 features broad commercial availability paired with enhanced software auditing protections, Mythos 5.1 remains restricted to credentialed partners enrolled in specialized verification initiatives.
Performance Benchmarks and Operational Efficiency
Performance evaluations published by Anthropic show significant capability gains across complex reasoning tasks and agentic workflows. In benchmark testing, Fable 5.1 achieved a 55.8% precision score on Terminal-Bench 4.0, while the restricted Mythos model surpassed 60% accuracy, according to company data. For scientific research conducted by autonomous agents via Terminal-Bench-Science 0.1, Fable 5.1 registered 52.6% precision—outperforming both the prior Fable 5 iteration at 24.7% and Opus 5 at 29%.
Furthermore, investment firm Millennium utilized Fable 5.1 during internal testing to successfully isolate the root cause of a rare system crash that had eluded human engineers for years, as noted in Anthropic documentation.
Token Pricing Structures and Cost Reductions
To support broader deployment, Anthropic restructured operational costs by maintaining a processing fee of $10 per million input tokens and $50 per million output tokens, as reported by company sources. Despite matching the headline rates of earlier high-end releases, the company optimized cache read pricing to deliver average savings of 25% on standard workloads. For complex agentic workloads involving continuous data reprocessing, operational costs drop by up to 45%.
Pro Tip: Enterprise customers utilizing Fable 5.1 can leverage cache read efficiencies to significantly lower expenses on long-running multi-turn agentic workflows and repository-scale code analysis.
Targeted Safeguards and Data Privacy Controls
Security architectures differ fundamentally between the two releases. Fable 5.1 introduces Enterprise Frontier Safeguards (EFS), a system designed to give enterprise users zero-retention data privacy by storing operational data entirely within customer-controlled cloud infrastructure rather than Anthropic’s environment, according to official statements. Additionally, Fable 5.1’s updated software auditing filters reduce false positives by 60% compared to previous versions.
Mythos 5.1 successfully achieved a 50% experimental success rate in designing protein binders across 12 evaluated biological targets and optimized GPU kernels for bioinformatic tasks with a 60% drop in inference costs.
Strict Access Control Policies for Mythos 5.1
Access to Claude Mythos 5.1 remains tightly regulated. Anthropic restricts usage strictly to participants enrolled in the Cyber Verification Program (CVP) and the Life Sciences Verification Program (LSVP), or legacy users with prior Project Glasswing approval. Fable 5.1 is broadly accessible today through the Claude API, web applications, and consumption-based enterprise tiers, while subscription-plan rollouts follow a phased schedule concluding in late June.
Frequently Asked Questions
What is the primary difference between Claude Fable 5.1 and Claude Mythos 5.1?
Both models share identical underlying capabilities, but Fable 5.1 includes commercial-grade safeguards and Enterprise Frontier Safeguards for general availability, while Mythos 5.1 operates with fewer restrictions and requires specialized partner verification.
How much do Fable 5.1 and Mythos 5.1 cost to use via API?
According to Anthropic’s pricing data, both models cost $10 per million input tokens and $50 per million output tokens, with cache read reductions offering up to 45% savings on complex agentic workloads.
How can developers access Claude Fable 5.1?
Fable 5.1 is available immediately via the Claude API using the model identifier claude-fable-5, as well as through consumption-based enterprise plans.
Join the Conversation: How do you plan to integrate advanced agentic workflows into your software development pipeline? Share your thoughts in the comments below, or subscribe to our newsletter for weekly artificial intelligence updates.
Related reading