Anthropic has launched Claude Fable 5, its most powerful Mythos-class model, which is safe and available for general use. Fable 5 is exceptional in terms of capabilities as compared to all generally available models. It outperforms across all benchmarks in knowledge work, vision, scientific research, and other areas. Alongside Fable 5, the company is also introducing Claude Mythos 5 for a group of cybersecurity professionals and infrastructure providers.
Anthropic claims that Mythos 5 has the strongest cybersecurity capabilities of any currently available model, though independent benchmarks have not confirmed this yet. Both these models can operate autonomously for longer than the previous Claude models.
What Can Claude Fable 5 Actually Do?
One key highlight of Claude Fable 5 is that it can handle extended, complex workloads. It supports a 1-million-token context window and generates up to 128,000 output tokens per request. In a 50-million-line Ruby codebase, the model performed codebase migration in a single day that would have taken a team two months.
Beyond agentic coding, Fable 5 performs well in knowledge-intensive tasks. It has achieved the highest scores in document interpretation, document-based reasoning, charts, and more.
Thereafter, Fable 5’s performance improvements from memory access were three times more than Opus 4.8’s improvements from the similar setup. The model also shows enhancements in visual understanding. According to Anthropic, Fable 5 can extract information from scientific diagrams, interpret complex charts, and recreate application source code from screenshots.
Claude Mythos 5: Powerful Model with Strict Security Controls
Claude Mythos 5 uses the same underlying model as Fable 5, but with some strong safety guards in some areas. It is only available through Project Glasswing, and in collaboration with the US government. The model is the successor of Mythos Preview and shares some capabilities without the safety classifier. Mythos 5 will expand soon through the Trusted Access Program.
Anthropic’s Commitment to Safety
Despite its standout capabilities, Anthropic has placed a strong emphasis on safety. Releasing the capable model comes with risks, particularly in biology, chemistry, cybersecurity, and other fields. To address these concerns, Fable 5 includes safety classifiers that track user requests.
Another noteworthy change following the launch is a 30-day data retention policy for traffic on Mythos-class models. Anthropic states, “We won’t use this data to train new Claude models, or for any non-safety-related purpose, and we’ve instituted new privacy protections including logging all human access to the data and ensuring its deletion after 30 days in almost all cases.” This policy applies specifically to the traffic processed by Mythos-class models including Fable 5 and Mythos 5.
In terms of pricing, both Claude Fable 5 and Claude Mythos 5 are available at $10 per million input tokens and $50 per million output tokens, making them less expensive than Anthropic’s Preview model.
Pricing and Availability
Claude Fable 5 is available publicly. However, Mythos is limited to Glasswing partners and will soon choose biology researchers with powerful biology and chemistry safeguards until its broader trusted access program is available.
Fable 5 is included on Pro, Max, Team, and seat-based Enterprise plans at no additional cost from June 9 to June 22. From June 23, Anthropic will remove Fable 5 from those plans. Usage credits will be required after this.
The joint launch marks a significant step by Anthropic in bringing advanced AI capabilities to the widest possible audience, with safety in mind.
To read the latest news happenings in the tech and innovation space, don’t miss out on visiting our website.
Recommended For You:
Claude Haiku 4.5 vs Claude Sonnet 4.5: Picking Your Perfect AI Partner