An article summarized by TechCrunch:
Anthropic has launched Claude Fable 5, the first public version of its powerful Mythos AI model, giving businesses and developers access to some of the company’s most advanced AI capabilities. Fable 5 is designed to excel at software engineering, knowledge work, analytics, and visual tasks, but Anthropic has built in strict safety guardrails. For high-risk topics such as cybersecurity, biology, chemistry, and model distillation, Fable automatically hands off requests to the less powerful Claude Opus 4.8 model. The launch follows months of limited testing with select partners and critical infrastructure organizations.
The release comes as Anthropic balances rapid AI advancement with growing safety concerns. The company previously restricted Mythos due to fears about cybersecurity risks and potential misuse. Before launching Fable 5, Anthropic conducted extensive security testing, including over 1,000 hours of jailbreak attempts and external red-team evaluations, claiming no universal exploits were discovered. To monitor potential threats, Anthropic is also requiring a 30-day data retention policy for all Fable 5 and Mythos 5 users, even for enterprise customers that previously had zero-retention agreements. The company says the data will not be used for training but instead for detecting attacks and improving safety systems.
Early performance results suggest Fable 5 could become one of the strongest AI models on the market. Analytics platform Hex said it was the first model to score above 90% on its benchmark for complex analytical tasks, while companies including Base44, Genspark, and Rakuten reported major improvements in coding, app development, tool usage, and autonomous workflows. However, that performance comes at a premium price: Fable 5 costs $10 per million input tokens and $50 per million output tokens, twice the cost of Claude Opus 4.8. Anthropic expects demand to be extremely high as it prepares for a potential IPO and competes with rivals like OpenAI in the race to build increasingly capable AI systems.
