On June 9, Anthropic released two versions of its Claude model family. Fable 5 is described as a Mythos-class solution that is safe for general use, while Claude Mythos 5 is a "private" base model with relaxed restrictions in certain areas.

The key difference between them lies in the operation of protective classifiers. The public model monitors sensitive queries in areas such as cybersecurity, biology, and chemistry, as well as attempts to distill the model. In such cases, the query is redirected to Claude Opus 4.8.

Source: Anthropic.

"To release the model safely and quickly, we configured these protective measures conservatively — sometimes they respond to harmless queries, although on average they trigger in less than 5% of sessions," the company emphasized.

Fable 5 is available through a standard subscription in the web interface and Claude API.

Both models are priced the same: $10 per million input tokens and $50 per million output tokens. The company noted that this is less than half the cost of Claude Mythos Preview.

More About the Closed Flagship

Claude Mythos 5 was opened to select trusted participants, including members of Project Glasswing. As part of the program, the new model replaced Claude Mythos Preview.

Anthropic plans to expand access for cybersecurity tasks and specific biomedical scenarios by the end of 2026.

Like Fable, Mythos has restrictions but offers expanded access in certain areas.

Anthropic has labeled the closed solution as its strongest model for cybersecurity tasks, which "excels at detecting and exploiting software vulnerabilities." Results in biology also show a tenfold acceleration in certain stages of drug development.

Source: Anthropic.

"Mythos 5 utilizes protein design and bioinformatics tools, but without human involvement, it matches or often surpasses experienced human operators. [...] Nine out of 14 protein targets in the study yielded promising compounds for drug development that we are currently exploring," the company stated.

For biomedical researchers, Anthropic is preparing a separate access pathway to the Fable 5 version without restrictions on biology and chemistry, but with censorship maintained in the cybersecurity domain.

The First Publicly Available Mythos-Class Model

Anthropic claims Fable 5 is the strongest of the Claude models released to the public, focusing on long and complex tasks. The company highlights areas such as programming, document analysis, vision, memory, and scientific research.

The startup has not disclosed engineering details about the model's architecture, including information on the number of parameters, training scheme, and stack description. An open license has also not been published.

During preliminary testing, Stripe reported that Fable 5 compressed months of development into just a few days. The solution migrated a Ruby codebase of 50 million lines in just one day, a task that would have taken the programming team two months, analysts noted.

Source: Anthropic.

Fable 5 is also more cost-effective in terms of token usage compared to previous Claude versions: in the FrontierCode evaluation by Cognition, the model shows superior performance among leading AI solutions.

Source: Anthropic.

In the Hebbia Finance Benchmark test for logical reasoning analysis, Fable 5 achieved the highest score among all models.

One of the model's key features is its "vision." The AI can extract precise numbers from detailed scientific diagrams and perform complex tasks, such as reconstructing the source code of a web application from just screenshots.

Additionally, Fable 5 requires fewer auxiliary tools. For example, previous Claude models struggled with playing Pokémon FireRed even with additional tools, but the new version completed it with minimal assistance.

https://www.youtube.com/watch?v=Ty_50J84fMY

Fable 5 has improved memory, maintaining focus on millions of tokens during long tasks and enhancing results by utilizing its own notes.

As a reminder, in early June, Anthropic representatives warned about the risks of AI self-improvement. According to internal data, over 80% of the code for the company's current products was written by Claude.