Anthropic has released Claude Fable 5.1, a new artificial intelligence model built specifically for enterprises tackling ambitious, data-intensive workflows that demand robust reasoning and autonomous task completion. The model represents a significant step forward in making AI practical for knowledge-intensive work like legal document review, financial analysis, scientific research, and large-scale software engineering projects.

What Makes Claude Fable 5.1 Different From Earlier Models?

Claude Fable 5.1 introduces several capabilities that set it apart from previous generations. The model can process up to 1 million input tokens, roughly equivalent to ingesting 100,000 words at once, allowing it to work with entire codebases, lengthy legal documents, or multi-stage research projects without losing context. This massive context window enables the model to maintain awareness of complex, long-running workflows that would overwhelm earlier AI systems.

The model also handles both text and images, meaning it can work directly with charts, diagrams, spreadsheets, architectural plans, and other visuals embedded in files or PDFs. This multimodal capability is particularly valuable for teams managing complex visual data alongside written information.

A feature called "adaptive thinking" runs continuously in the background, allowing Claude Fable 5.1 to plan multi-step tasks, detect its own mistakes, and refine outputs without constant human guidance. The model uses proactive tool use, progress tracking, and error recovery to navigate workflows autonomously.

How Can Enterprises Use Claude Fable 5.1 for Their Operations?

• Legal and Financial Teams: Organizations handling document-heavy workflows in law, finance, or compliance can use the model to review contracts, analyze regulatory filings, and process large datasets while maintaining full visibility into how the AI reached its conclusions.
• Software Engineering Leaders: Development teams working on large codebases can leverage the model for code reviews, performance optimizations, and autonomous testing across interdependent modules, with the model's vision capabilities enabling it to validate changes against architectural diagrams.
• Project Managers and Knowledge Workers: Teams assigning agentic tasks can rely on Claude Fable 5.1 to actively plan, execute across tools, recover from errors, and provide progress updates without continuous oversight, freeing human managers to focus on strategic decisions.
• Decision-Makers in Regulated Industries: Executives in healthcare, finance, and government sectors benefit from the model's enhanced safety features, including automatic fallback to less capable models for high-risk requests such as bioweapons or exploit code generation, ensuring compliance with regulatory requirements.

What Are the Pricing and Safety Features?

Anthropic has structured pricing to make the model economical for heavy users. Input tokens cost $10 per million, while output tokens cost $50 per million. The most significant cost reduction comes through cache reads, which now cost just $0.25 per million tokens, approximately 75% less than previous pricing. This dramatic reduction in cache costs makes it practical for organizations that repeatedly reference the same documents or codebases.

For organizations concerned about data location, Fable 5.1 supports inference restricted to US-only regions for about 10% higher cost. This addresses compliance requirements for companies handling sensitive government or regulated industry data.

Safety features include content provenance tracking, making it visible where information originated, and automatic fallback mechanisms that route certain high-risk requests to less capable models. These safeguards make the model attractive for use cases involving sensitive data or strict compliance requirements.

Who Is Claude Mythos 5.1, and How Does It Relate to Fable?

Anthropic also released Claude Mythos 5.1, which shares identical features and specifications with Fable 5.1 but operates under restricted access through Anthropic's Project Glasswing program. The two models are functionally equivalent, with Mythos availability limited to invited users. Both models support the same 1 million token context window, multimodal input, and pricing structure.

The release of these higher-tier models comes as Anthropic continues to expand its model lineup to serve different enterprise needs. Earlier this year, Anthropic published a threat intelligence report documenting how its Claude models were being misused by state-sponsored groups, financially motivated criminals, and commercial spyware vendors. The company has since implemented additional safeguards, including requiring accounts operating from unsupported countries to verify their identity or lose access.

For organizations wrestling with ambitious, data-intensive, or long-duration workflows that require robust reasoning, vision capabilities, and agentic autonomy, Claude Fable 5.1 represents one of the most capable options available in the current AI landscape. However, the model's higher pricing means that for simpler or shorter tasks, less capable and less expensive models may be more economical.