Anthropic unveils Claude Fable 5, limits access to advanced Mythos model

The AI firm says Fable 5 sets a new benchmark across tasks, while access to Mythos 5 remains limited due to concerns over its advanced cybersecurity capabilities

Anthropic launched Claude Fable 5 on Tuesday, its latest Mythos-class model, which the frontier artificial intelligence (AI) company says is superior across all tasks. Access to its Mythos 5 model, however, remains restricted to a handful of users because of its advanced capabilities in detecting software flaws.

 

Fable 5 is state-of-the-art on nearly all tested benchmarks, the company said, with strong performance in software engineering, knowledge work, scientific research, and vision tasks. “Releasing a model this capable comes with risks. Without safeguards, Fable 5’s capabilities in areas like cybersecurity could be misused to cause serious damage. Queries on a narrow range of topics will instead receive a response from our next-most-capable model, Opus 4.8,” Anthropic posted on X.

 

The launch comes as Anthropic — valued at nearly $1 trillion — and its closest competitor, OpenAI, have confidentially filed for initial public offerings with the US market regulator, a move expected to attract strong investor interest. “At present, it seems to be a better version of the Opus model,” said Pankit Desai, cofounder and chief executive officer of cybersecurity firm Sequretek.

 

“Its agentic AI capabilities are stronger, and its ability to build code through vibe coding is also much better. However, the pricing is almost double that of Opus. It will take time for enterprises to warm up to it considering if you have a certain model, you will be resistant to change,” he added.

 

The company has also released Claude Mythos 5, a separate version of the same system that remains restricted to vetted organisations under its Project Glasswing programme.

 

The dual release marks a shift in how Anthropic is deploying frontier AI systems. Rather than limiting access entirely, the company is separating capability from access, making a controlled version publicly available while keeping the full model within restricted environments.

 

Anthropic said it intends to expand access to Mythos 5 through a broader trusted-access programme for defensive cybersecurity work and biomedical research. Pricing for both models is set at $10 per million input tokens and $50 per million output tokens.