Anthropic Launches Claude Fable 5, Its Most Powerful AI Model, with Robust Safety Measures

Anthropic has officially made its most powerful artificial intelligence model, Claude Fable 5, accessible to the general public for the first time. This release marks a significant milestone, as Fable 5 is the first publicly available version of the company's highly advanced Mythos model. Anthropic highlights Fable 5's exceptional capabilities in software engineering, knowledge work, and vision tasks, positioning it as a versatile tool for various applications. However, this power is balanced with stringent safety limits, particularly in high-risk domains such as cybersecurity, biology, chemistry, and distillation, where the model is programmed to block responses and defer to the less potent Claude Opus 4.8. The journey to public availability for this technology has been carefully managed. Initially, the Mythos model was launched as a preview in April, with access restricted to a select group of partners due to cybersecurity concerns. This access was later expanded to hundreds of organizations across 15 countries, primarily focusing on entities managing critical infrastructure. Now, a version of this cutting-edge technology is available to anyone through Anthropic's Claude API and consumption-based Enterprise plans. The rollout for subscriptions will occur in stages, with Fable 5 included in Pro, Max, Team, and seat-based Enterprise plans at no extra cost until June 22, after which usage credits will be required, with plans to restore it as a standard feature as soon as possible. Concurrently, Anthropic is deploying Mythos 5, an even newer version, to organizations already approved for the advanced model. This launch is strategically timed as Anthropic prepares for its public market debut, joining other tech giants like OpenAI and SpaceX. It also follows Anthropic's recent public plea for major global AI labs to establish a coordinated 'brake pedal' on frontier AI development. The company has voiced concerns about the rapid advancement of AI systems, warning that they may soon achieve recursive self-improvement (RSI), where models autonomously enhance themselves without human intervention. To mitigate risks, Anthropic rigorously stress-tested Fable 5's classifiers against numerous 'jailbreak' attempts, conducting over 1,000 hours of internal testing and collaborating with external red-teaming organizations, none of which found universal jailbreaks. Despite these extensive safety measures, Anthropic acknowledges the potential for novel attacks. Consequently, with the introduction of Fable 5 and Mythos 5, the company is implementing a mandatory 30-day data retention policy for all traffic, even for enterprises that previously had zero-retention agreements. Anthropic clarifies that this data will not be used for training purposes but solely to "defend against complex and novel attacks, including new jailbreaks," and to "identify and reduce false positives." This policy could set a significant industry precedent, where access to increasingly powerful AI models becomes contingent upon mandatory data retention framed as a critical safety measure. While Fable 5 offers unparalleled capabilities, its pricing reflects its advanced nature, set at $10 per million input tokens and $50 per million output tokens – double the cost of Opus 4.8. This pricing strategy might serve as a deterrent for some, especially as many enterprises are becoming more critical of escalating AI costs. However, third-party evaluations underscore Fable 5's value. Analytics company Hex reported that Fable was the first model to achieve a 90% score on its core analytics benchmark for complex, long-running tasks, noting its "strong judgment and attention to nuance." Similarly, Base44 praised Fable's ability to "one-shot full apps" and its excellent tool-calling, while Genspark found it outperformed every other model in UI design and game coding. Companies like Rakuten believe the "extra thinking pays for itself," enabling highly autonomous operations.