Anthropic Launches Claude Fable 5

Anthropic Launches Claude Fable 5

Anibel Adjei
June 10, 2026
Tech, AI

Anthropic Launches Claude Fable 5, Bringing Mythos-Class AI to the Public With New Safety Guardrails

SAN FRANCISCO, June 10, 2026, Anthropic has unveiled Claude Fable 5, the first publicly available AI model built on the company’s powerful Mythos-class architecture, marking a significant milestone in the race to develop more capable artificial intelligence systems.

The company describes Fable 5 as its most advanced publicly accessible model to date, delivering state-of-the-art performance across software engineering, knowledge work, scientific research, vision tasks, and long-running autonomous workflows. According to Anthropic, the model outperforms previous Claude generations by a substantial margin, particularly on complex and extended tasks.

Alongside Fable 5, Anthropic also announced Claude Mythos 5, a more powerful version reserved for approved cybersecurity organizations, critical infrastructure providers, and selected research partners through the company’s Project Glasswing initiative. Mythos 5 operates with fewer restrictions and is intended for defensive cybersecurity and advanced scientific research.

A Powerful Model With Built-In Limits with Claude Fable 5

Unlike previous Claude releases, Fable 5 includes a new layer of safety protections designed to prevent misuse in high-risk areas such as cybersecurity, biology, chemistry, and AI model distillation.

When the system detects requests in these sensitive categories, it automatically redirects them to Claude Opus 4.8, a less capable but safer model. Anthropic says more than 95% of user sessions never trigger this fallback mechanism, allowing most users to experience the full capabilities of Fable 5.

The company says these safeguards were necessary because Mythos-class models have reached a capability level where they could potentially assist malicious actors if deployed without restrictions.

Image Source and Credits: Anthropic

Major Gains in Coding and Research

Anthropic highlighted several real-world tests demonstrating the model’s capabilities.

Financial technology company Stripe reported that Fable 5 completed a codebase migration in a 50-million-line Ruby project within a single day, a task that would typically require a team of engineers working for more than two months.

The model also achieved strong performance in analytical reasoning, document interpretation, and scientific research. In vision-related tasks, Fable 5 can reconstruct application source code directly from screenshots and extract detailed information from scientific diagrams.

Meanwhile, Mythos 5 demonstrated impressive results in biology and drug discovery research. Anthropic claims the model accelerated parts of the drug-design process by roughly ten times and generated novel scientific hypotheses that researchers preferred over those produced by previous-generation models.

New Security Measures Introduced

Anthropic is also implementing stricter security policies for Mythos-class models.

The company will require a mandatory 30-day data retention period for all traffic involving Fable 5 and Mythos 5. Anthropic says the retained data will not be used to train future AI models but will instead help identify jailbreak attempts, emerging threats, and weaknesses in its safety systems.

According to the company, extensive internal testing and external red-team exercises failed to discover any universal jailbreak capable of bypassing the model’s safeguards.

Pricing and Availability

Claude Fable 5 is available immediately through Anthropic’s API and enterprise offerings. The model is priced at $10 per million input tokens and $50 per million output tokens. Anthropic says the pricing is less than half that of the earlier Mythos Preview program.

For subscription users, Fable 5 will be included in Pro, Max, Team, and seat-based Enterprise plans through June 22. After that date, access will transition to a usage-credit model until Anthropic expands computing capacity.

The launch signals a new phase in the AI industry, where companies are attempting to balance increasingly powerful models with stronger safeguards. As frontier AI capabilities continue to advance, Anthropic’s approach suggests that future systems may become more capable while simultaneously operating under tighter security controls.

Written by Anibel

Read More from Tech Category

Leave a ReplyCancel Reply