Anthropic expands AI cybersecurity models with Claude Fable 5 and Mythos 5

Anthropic has made headlines by releasing its most advanced Artificial Intelligence (AI) model to the public, albeit with a focus on safety and responsible usage. The company has launched Claude Fable 5, touted as the first publicly accessible version of its cybersecurity-focused model, Mythos. In addition to Claude Fable 5, Anthropic has also rolled out Claude Mythos 5, which operates on the same underlying technology. According to the company, Claude Fable 5 is described as 'a Mythos-level model tailored for your most ambitious, long-term projects.' This model is engineered to handle demanding tasks that necessitate advanced reasoning, planning, creativity, and expertise. Such tasks include writing comprehensive texts, developing software, conducting thorough research, and managing intricate workflows. One of the standout features of Claude Fable 5 is its capability to assist over prolonged periods. It can remember project details across multi-step tasks that may span days, weeks, or even months. The AI is designed to plan across various stages, delegate tasks to sub-agents, and verify its own outputs, functioning like a reliable teammate capable of independently managing extensive, research-intensive projects and producing near-final results autonomously. Moreover, Claude Fable 5 can interpret and understand visual data, such as diagrams, charts, and tables present in documents and PDFs. For a limited time until June 22, users on Pro, Max, Team, and enterprise subscriptions can access Claude Fable 5 without incurring additional costs. However, after this period, users will need to pay based on their usage, though the company assures that this pricing change is temporary, with plans to reintegrate Fable 5 into regular subscriptions. Anthropic also announced the launch of Claude Mythos 5, a more potent and less restrictive iteration of Fable 5, specifically designed for cybersecurity professionals. Initially, this model will be available to a select group associated with the US government and partners under Project Glasswing, with broader access expected for trusted users in the future. Mythos 5 boasts enhanced capabilities in cybersecurity tasks, such as identifying vulnerabilities, analyzing threats, supporting defensive security measures, and aiding in cyber incident responses. In terms of pricing, both Fable 5 and Mythos 5 will cost $10 for every million input tokens and $50 for every million output tokens, which is significantly less than half the cost of the previous Claude Mythos Preview.