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

Anthropic has opened access to its most powerful and capable Artificial Intelligence (AI) model to the public, but it is doing so with caution by putting safety controls and usage restrictions. The AI company has launched Claude Fable 5, which is said to be the first publicly available version of its cybersecurity model, Mythos. In addition, the company has also introduced Claude Mythos 5 for cybersecurity, which uses the same core technology as Claude Fable 5

Claude Fable 5

Anthropic claims that Claude Fable 5 is “a Mythos-level model built for your most ambitious, long-running projects.” The model is designed to run demanding tasks that require deeper reasoning, planning, creativity, or expertise, such as writing a book, developing software, conducting extensive research, or managing complicated workflows.

Must read: AI building AI? Anthropic warns future is closer than you think

It also comes with the ability to assist over extended periods, as it can remember details through multi-step projects that run for days, weeks, or even months. The company said it can “plan across stages, delegating to sub-agents, and checking its own work.”

Claude Fable 5 can conduct complex, multi-stage work without constant supervision. It can work like an AI teammate that can independently manage large, research-heavy projects, complete the majority of work, and deliver near-finished results autonomously. It can also understand and interpret visual elements like diagrams, charts, and tables in files and PDFs. 

Claude Fable 5 availability

Until June 22, the Claude Fable 5 will be available at no additional cost to users on Pro, Max, Team, and enterprise subscriptions. After that, users who want to use the model will have to spend extra based on their usage. The company says this change is temporary and that it intends to bring Fable 5 back as a regular feature included in subscriptions.

Must read: Anthropic expands Claude Mythos access to India; Adds 15 other countries and 150 organisations

Claude Mythos 5

Anthropic also introduced Claude Mythos 5, which is a more powerful and less restricted version of Fable 5 designed for cybersecurity experts. The model will be initially rolled out to a selected group for the US government and partners under Project Glasswing. The company has also planned broader access later to trusted users. 

Mythos 5 is said to be an upgraded version of Claude Mythos Preview. The company claims that Mythos 5 comes with powerful cybersecurity capabilities, and it can conduct tasks such as identifying security vulnerabilities, analysing threats, assisting with defensive security operations, and supporting cyber incident response.

Fable 5 and Mythos 5 cost $10 for every million input tokens and $50 for every million output tokens, which is less than half the price of Claude Mythos Preview.