Anthropic Launches 'Claude Fable 5' and 'Mythos 5': Features, Price, and Differences Between the Two

Liputan6.com, Jakarta - Anthropic officially launched two of its newest artificial intelligence (AI) models, Claude Fable 5 and Claude Mythos 5, on June 9, 2026.

These two models mark the debut of Anthropic's most advanced 'Mythos' family, placing them at the top of the company's AI technology hierarchy, surpassing the existing Opus class.

While both share the same underlying model, the key difference lies in the implementation of safeguards tailored to specific use cases.

Claude Fable 5 is generally available, while Claude Mythos 5 is designed for restricted access to sensitive domains.

Claude Fable 5 is publicly accessible through Amazon Bedrock, the Claude Platform on AWS, and the Claude API.

Claude Mythos 5 is initially available only to a small group of verified partners, including those involved in Project Glasswing, which focuses on cybersecurity and infrastructure providers, and select life science researchers.

Anthropic plans to expand access to Mythos 5 through a broader trusted access program in the future.

Introducing Claude Fable 5: a Mythos-class model that we’ve made safe for general use.Its capabilities exceed those of any model we’ve ever made generally available. pic.twitter.com/2AvmEjHIX8

— Claude (@claudeai) June 9, 2026

The Fable 5 and Mythos 5 models are claimed to be leaders in nearly all tested AI benchmarks.

A key capability that stands out is their capacity for long-term, asynchronous execution, enabling them to handle complex tasks such as software engineering and knowledge-intensive work without requiring continuous intervention.

Both models also feature advanced vision capabilities that enable them to understand diagrams, graphs, and tables stored in various file formats, including PDF.

This feature opens up significant opportunities for document-intensive research and analysis in the finance, legal, analytics, architecture, and even game development sectors.

In coding contexts, these models can implement high-fidelity designs and use vision to critique their output against established objectives.

These models are capable of updating their own skills based on learning, developing their own tools and evaluations, and autonomously verifying their work.

While using the same base model, Claude Fable 5 and Claude Mythos 5 are differentiated by their safeguard configurations.

Claude Fable 5 is equipped with strong safeguards to limit its performance in high-risk areas such as cybersecurity, biology, and chemistry.

Requests deemed dangerous or sensitive in these domains are redirected to receive responses from less capable models, such as Claude Opus 4.8.Conversely, Claude Mythos 5 is the same base model as Fable 5, but with safeguards lifted in some areas, particularly for cybersecurity and life sciences capabilities.

In software engineering, Fable 5 scored 80.3% on SWE-Bench Pro, outperforming Claude Opus 4.8 (69.2%) and GPT-5.5 (58.6%).

As a real-world example, Stripe reported that Fable 5 successfully completed a 50 million-line codebase migration in a single day, a task that would have previously taken the full team over two months.

For scientific research, Mythos 5 is the first Anthropic model to consistently generate novel and compelling scientific hypotheses.

In cybersecurity, Mythos 5 is said to have the strongest capabilities of any model in the world, capable of identifying vulnerabilities in critical infrastructure.

Claude Fable 5 and Claude Mythos 5 are offered at a uniform price of $10 per million input tokens and $50 per million output tokens.

This price is about double that of the most advanced Claude Opus 4.8 model, but less than half the price of Claude Mythos Preview.

Anthropic and early customers report that Fable 5 is able to complete tasks with fewer turns and token consumption.

To further improve cost efficiency, Anthropic also offers a 90% discount on input tokens with its prompt caching feature.