Model Release

Anthropic Launches Claude Fable 5, Its Most Powerful Model

Anthropic released Claude Fable 5, its most powerful model, on June 9, 2026, with safety measures to limit high-risk applications.

A robotic dog navigates an indoor setting amidst red chairs, showcasing technology in modern environments.

Photo: Vladimir Srajber / Pexels

Anthropic has released Claude Fable 5, its most powerful AI model to date, to the general public. The model is part of the broader Mythos series and comes with safety measures to prevent misuse in high-risk areas. Fable 5 excels in software engineering, knowledge work, and vision tasks but is restricted in domains like cybersecurity, biology, and chemistry. In these areas, the model blocks responses and defaults to Claude Opus 4.8. The release follows a phased rollout, with access expanding to hundreds of organizations across 15 countries. Access through the Claude API and Enterprise plans is available through June 22, after which usage credits will be required. Anthropic is also deploying a new version of Mythos, called Mythos 5, to approved organizations.

The launch of Fable 5 coincides with Anthropic's preparations to enter the public markets, alongside OpenAI and SpaceX. The company has also called for a coordinated brake on frontier AI development due to concerns about rapid advancements leading to recursive self-improvement. To ensure safety, Anthropic conducted extensive testing, including an external bug bounty and collaboration with red-teaming organizations, which found no universal jailbreaks. Despite this, the company acknowledges the possibility of novel attacks and has implemented a 30-day data-retention policy for all traffic. The policy aims to defend against complex attacks and reduce false positives, potentially setting an industry standard for safety measures.

Anthropic emphasized that Fable 5's limitations are rare, with early data showing at least 95% of sessions running entirely on the model. Third-party testing highlighted Fable 5's performance, with Hex noting it achieved a 90% score on its analytics benchmark. Base44 praised its ability to create full apps with one-shot coding, while Genspark reported it outperformed other models in UI design and game coding. Pricing for Fable 5 and Mythos 5 is $10 per million input tokens and $50 per million output tokens, double that of Opus 4.8. This pricing may deter widespread adoption, especially as enterprises face rising AI costs.

Source: techcrunch

Key points

Anthropic released Claude Fable 5, its most powerful model, on June 9, 2026.
Fable 5 excels at software engineering, knowledge work, and vision tasks but is restricted in high-risk areas like cybersecurity, biology, and chemistry.
The model blocks responses in high-risk areas and defaults to Claude Opus 4.8.
Anthropic conducted extensive testing, including an external bug bounty and collaboration with red-teaming organizations, which found no universal jailbreaks.
The company implemented a 30-day data-retention policy for all traffic to defend against complex attacks and reduce false positives.
Third-party testing showed Fable 5 achieved a 90% score on its analytics benchmark.
Pricing for Fable 5 and Mythos 5 is $10 per million input tokens and $50 per million output tokens, double that of Opus 4.8.

Source: TechCrunch Read the original →

WRITTEN BY

Alex Lindgren

LLMs & Frontier Models

Alex covers the large language models and their impact on society.