Anthropic, a leading force in artificial intelligence, has just opened the gates to its most advanced AI model for the general public. This marks a significant milestone, as users can now access a version of their formidable Mythos model, known as Claude Fable 5.
Launched this week, Fable 5 is designed to excel in demanding areas such as software engineering, complex knowledge work, and sophisticated vision tasks. However, this powerful new tool arrives with a strong emphasis on safety, incorporating strict guardrails to prevent misuse in high-risk domains.
Introducing Claude Fable 5: Accessible Power
Fable 5 represents the first publicly available iteration of Anthropic’s cutting-edge Mythos technology, a model that previously saw a limited preview in April due to cybersecurity considerations. Initially, access was restricted to a select group of partners before expanding to hundreds of organizations across 15 countries, primarily those managing critical infrastructure.
Now, this advanced technology is becoming more broadly available, initially through Anthropic’s Claude API and consumption-based Enterprise plans. For subscribers, the rollout will occur in phases, with Fable 5 included at no extra cost in Pro, Max, Team, and seat-based Enterprise plans until June 22.
Following this initial period, Anthropic will temporarily require usage credits for Fable 5 from June 23, with plans to reintroduce it as a standard subscription feature as soon as possible. Concurrently, Anthropic is also deploying an even newer version, Mythos 5, to its pre-approved organizational partners.
Balancing Innovation with Robust Safety Measures
The public launch of Fable 5 comes at a pivotal time for Anthropic, as the company prepares for its potential entry into public markets. This move also aligns with their broader advocacy for a “coordinated brake pedal” on frontier AI development, driven by concerns that rapidly advancing systems could soon achieve Recursive Self-Improvement (RSI) – autonomously improving themselves without human intervention.
Mindful of the immense capabilities of a Mythos-class model, Anthropic has undertaken extensive stress-testing to ensure Fable 5’s safety. They conducted an internal bug bounty, which saw no universal jailbreaks discovered across more than 1,000 hours of testing, and further collaborated with external red-teaming organizations who similarly failed to bypass the safety mechanisms.
Despite these rigorous efforts, Anthropic acknowledges that novel attacks may still emerge. Therefore, a new policy accompanying Fable 5 and Mythos 5 mandates a 30-day retention on all traffic, even for enterprises that previously held zero-retention agreements. This data will be used solely to defend against complex and novel attacks, including new jailbreaks, and to reduce false positives, emphatically not for model training. This policy, a potential industry precedent, mandates data retention as a safety measure for powerful AI access.
Performance, Pricing, and the Enterprise Outlook
While Fable 5 boasts impressive capabilities, Anthropic states that in rare, high-risk scenarios, the model will defer to Claude Opus 4.8. Early data indicates this happens infrequently, with at least 95% of Fable sessions running entirely on the model’s own responses, showcasing its robust independent performance.
Third-party evaluations have underscored Fable 5’s exceptional abilities. Analytics firm Hex reported that Fable was the first model to achieve a 90% score on its core analytics benchmark, highlighting its strong judgment and attention to nuance in complex analytical tasks. Similarly, vibe-coding platform Base44 praised Fable’s capability for “one-shotting full apps” and its excellent tool-calling, while Genspark noted its superior performance in UI design and game coding compared to other models.
However, this advanced capability comes with a higher price tag. Both Fable 5 and Mythos 5 are priced at $10 per million input tokens and $50 per million output tokens, effectively doubling the cost of Opus 4.8. This pricing structure could serve as a significant consideration for enterprises, many of whom are already scrutinizing AI expenditures after encountering unexpected bills or blowing through their annual budgets.
Despite the increased cost, some enterprises recognize the immense value. Shopping rewards platform Rakuten, for instance, finds Fable 5’s ability to reflect on and validate its own work invaluable. They state that “the extra thinking pays for itself,” enabling highly autonomous operations and justifying the investment for critical business functions.
The public debut of Claude Fable 5 marks a pivotal moment in the accessibility of frontier AI, offering unprecedented power for a range of complex tasks. Anthropic’s dual commitment to innovation and stringent safety protocols, even with new data retention policies, defines its approach to responsible AI deployment.
As enterprises continue to navigate the evolving landscape of AI adoption, Fable 5 presents a compelling option, balancing cutting-edge performance with the vital guardrails necessary for powerful technology. Its impact on software development, knowledge work, and AI-driven solutions is poised to be substantial.
Source: TechCrunch – AI
