Anthropic launches Fable 5 and Mythos 5: What’s new in AI models

Anthropic has officially released two groundbreaking AI models today: Claude Fable 5 and Claude Mythos 5, marking a significant leap in artificial intelligence capabilities. These models represent the company’s first broad deployment of its advanced Mythos-class technology, previously restricted to select partners in its Project Glasswing cybersecurity initiative.

Fable 5 is positioned as the most powerful generally available model Anthropic has ever offered. It outperforms all prior Claude models across critical benchmarks, excelling in software engineering, knowledge work, vision tasks, scientific research, and long-running operations. While the prior Claude Mythos Preview still leads in specialized areas like computer use and multidisciplinary reasoning, Fable 5 delivers superior versatility and accessibility for everyday users.

Key distinctions: Fable 5 vs. Mythos 5

The primary difference between these models lies in their availability and safeguards, not their core capabilities. Here’s how they compare:

• Fable 5 is designed for mass adoption, integrating Mythos-class intelligence with enhanced safety protocols. Requests deemed high-risk—such as those involving cybersecurity, biology, chemistry, or model distillation—are automatically rerouted to Claude Opus 4.8, Anthropic’s flagship general-purpose model. Users receive clear notifications when this occurs.

• Mythos 5, by contrast, retains the same underlying capability but with fewer restrictions. It is currently accessible only to Anthropic-approved users, including cybersecurity partners in Project Glasswing and select biology researchers. This model omits certain safeguards present in Fable 5, allowing for broader functionality in controlled environments.

Anthropic emphasizes that Fable 5 maintains robust security without compromising performance. Internal and external red-teaming tests—totaling over 1,000 hours—revealed no universal jailbreaks, and the company reports that 95% of Fable 5 sessions operate independently, without fallback mechanisms.

Pricing and rollout strategy

Both models are priced at $10 per million input tokens and $50 per million output tokens, positioning them as the most expensive among major AI models globally. While this cost is less than half the price of the previous Claude Mythos Preview, it remains significantly higher than competitors like DeepSeek v4 Flash ($0.42 total) or Gemini 3.1 Flash-Lite ($1.75 total).

For developers, Fable 5 is immediately available via the Claude API under the identifier claude-fable-5. It is also fully supported on consumption-based Enterprise plans.

The rollout for subscription users follows a staggered approach:

• From today through June 22, Fable 5 will be included at no extra cost on Pro, Max, Team, and seat-based Enterprise plans.

• Starting June 23, accessing Fable 5 will require usage credits, though Anthropic has pledged to reinstate it as a standard feature in subscription plans as soon as possible.

Mythos 5, meanwhile, remains invitation-only for existing users of the Claude Mythos Preview. Anthropic has not announced plans to expand its availability beyond its current restricted user base.

Why these models matter for the future of AI

The launch of Fable 5 and Mythos 5 underscores Anthropic’s commitment to balancing power, safety, and accessibility in AI development. By offering a high-performance model with built-in safeguards for the public while reserving unrestricted capabilities for specialized use cases, the company is setting a new standard for responsible AI deployment.

As AI continues to evolve, these models could redefine expectations for enterprise applications, research acceleration, and long-form content generation. Whether Fable 5’s premium pricing will deter widespread adoption or drive demand for its unparalleled capabilities remains to be seen—but one thing is clear: the AI landscape has just become more competitive.