#What are the features of Claude Fable 5?
Claude Fable 5 represents a significant development in AI technology. Released by Anthropic, this model is the most advanced of its kind to become widely available. It excels in various fields, including software engineering, knowledge work, vision, scientific research, and other intricate tasks. This model is particularly effective for lengthy and complex assignments, marking a notable enhancement from previous versions of the Claude system.
#How does Claude Fable 5 address safety?
With increased capabilities comes the need for enhanced safeguards. Claude Fable 5 is designed to ensure that when it encounters requests linked to sensitive areas such as cybersecurity or biology, it will redirect those inquiries to Claude Opus 4.8 for evaluation. This precautionary measure aims to mitigate risks associated with the model’s more advanced functionalities. Nevertheless, Anthropic has stated that these safeguards are purposefully conservative. They expect the safeguards to activate in fewer than 5% of sessions, maintaining efficiency while ensuring security.
#What is Claude Mythos 5 and its applications?
Accompanying Fable 5 is Claude Mythos 5, which retains the underlying technology but is tailored for a designated group of cybersecurity professionals and infrastructure providers. This model will initially roll out under Project Glasswing, collaborating with the US government as an upgrade to the Claude Mythos Preview. This initiative boasts unparalleled cybersecurity capabilities in the AI landscape and is set to widen access for both cybersecurity entities and those in the biological research fields.
#What are the performance highlights of Fable 5?
Early evaluations of Claude Fable 5 have highlighted its superior performance in software engineering. For instance, a major company reported that the AI efficiently accomplished a code migration task within a massive 50 million line Ruby codebase in just one day. Such a feat would typically demand a team months to complete. The model has also demonstrated impressive abilities in agentic coding, financial reasoning, and interpreting complex charts, proving suitable for sophisticated vision-related tasks.
In addition, Fable 5 exhibits enhanced long-context handling and memory, allowing it to maintain focus across extensive tasks while refining its outputs by utilizing its notes. This improvement positions Fable 5 as an ideal option for autonomous operations over extended periods.
#How is Mythos 5 reshaping life sciences research?
Mythos 5 also serves as a research tool within the life sciences domain. Reports indicate that it has substantially expedited elements of the drug design process for internal protein design experts, accomplishing tasks ten times faster than traditional methods. Moreover, assessments have shown that its molecular biology hypotheses surpass those generated by Opus class models, as evidenced by blinded comparisons.
#What are the new policies related to data retention?
The launch of these models introduces a new data retention policy. Anthropic mandates a 30-day retention period for data traffic associated with both Fable 5 and Mythos 5, as well as future models of comparable or greater capability. This data will serve to monitor safety, ensuring that it is not repurposed for training new Claude models.
#How can you access and utilize Claude Fable 5?
Claude Fable 5 is accessible via the Claude API and available to business customers on a consumption-based enterprise model. It is included at no additional cost in Pro, Max, Team, and seat-based enterprise plans until June 22. After this date, it will transition to a usage credit system unless current capacity allows for an extension. Both Fable 5 and Mythos 5 are offered at competitive rates of $10 per million input tokens and $50 per million output tokens, significantly lower than the previous Claude Mythos Preview, and demand is anticipated to surge. As capacity increases, wider subscription access is expected to re-emerge.