Anthropic Releases Claude Fable 5, Its Most Advanced Mythos-Class AI Model

By: Harsh Vardhan

|

Updated on: 10-Jun-2026 12:00 PM

5,676 views

Follow Us:

5,676 views

Anthropic has launched Claude Fable 5, its first publicly available Mythos-class AI model, marking a significant step in advanced artificial intelligence. The company announced on Tuesday that Fable 5 is now accessible, but with strict safeguards in place for high-risk areas.

Key Highlights

• Anthropic launches Claude Fable 5, its first publicly available Mythos-class AI model.
• Fable 5 outperforms previous models and includes strict safeguards for high-risk queries.
• Pricing set at $10 per million input tokens and $50 per million output tokens.

Claude Fable 5 Capabilities and Safeguards

Claude Fable 5 is Anthropic’s most capable generally available model to date. It excels in software engineering, knowledge work, vision, scientific research, and handling long, complex tasks. According to Anthropic, Fable 5 outperforms previous models, scoring over 10 percent higher than Claude Opus 4.8 on some benchmarks.

Anthropic has implemented hard limits on Fable 5 in sensitive domains such as cybersecurity, biology, chemistry, and model distillation. When users attempt high-risk queries, the model blocks the response and defaults to Claude Opus 4.8. These measures aim to address global concerns about the potential misuse of powerful AI, especially in cybersecurity.

Dianne Penn, Anthropic's head of product management for research, stated that the broad release was possible due to new safety guardrails. She explained that the company designed new classifiers and safeguards to ensure the technology provides more benefits than harm. The model blocks responses to questions about dangerous topics, such as instructions for making hazardous substances.

Anthropic tuned these safeguards conservatively, so the fallback to Opus 4.8 may occur more often than some users expect. However, early data shows that at least 95 percent of Fable 5 sessions run entirely on the model’s own responses.

Testing, Availability, and Pricing

Before release, Anthropic stress-tested Fable 5’s safety features with over 1,000 hours of internal bug bounty testing and external red-teaming. No universal jailbreaks were found during these tests. The company also worked with outside organizations to further evaluate the model’s security.

Alongside Fable 5, Anthropic introduced Claude Mythos 5 for organizations approved for advanced access. Mythos 5 uses the same underlying model as Fable 5 but with some safeguards lifted. Access to Mythos 5 remains limited to select partners, including cyber defenders and critical infrastructure providers through Project Glasswing.

Anthropic requires 30-day retention on all traffic for Fable 5 and Mythos 5, including for enterprises with previous zero-retention agreements. The company states it will not use this data for training but only to defend against complex attacks and new jailbreaks.

Claude Fable 5 is available through the Claude API and consumption-based Enterprise plans. Access for subscription plans will roll out in stages. Until June 22, Fable 5 is included in Pro, Max, Team, and seat-based Enterprise plans at no extra cost. After June 23, usage credits will be required, with plans to restore standard access when capacity allows.

Pricing for both Fable 5 and Mythos 5 is set at $10 per million input tokens and $50 per million output tokens. This is double the price of Opus 4.8. Anthropic expects high demand for Fable 5 and notes that availability may be limited as a result.