Anthropic, the developer behind the Claude artificial intelligence models, has released a new version of its technology to the general public while imposing restrictions on its use in sensitive domains. The model, named Fable 5, is the first widely available offering from the company's Mythos class, its most advanced AI lineup that was unveiled in April but previously limited to a small group of partner institutions due to cybersecurity concerns.
Features and Availability
Fable 5 is designed for tasks such as writing and debugging software code, answering complex research questions, and analyzing images. Alongside this public release, Anthropic is providing an unrestricted version, Claude Mythos 5, to organizations already enrolled in its Project Glasswing program, which includes cybersecurity partners. This select group expanded in early June to approximately 200 organizations across more than 15 countries, with further growth anticipated.
Security Measures
Anthropic had restricted access to the Mythos models due to their potential to quickly identify vulnerabilities in critical infrastructure, including banking platforms and power grids. When Project Glasswing launched, some critics accused the company of overhyping the threat. However, companies that have tested Mythos have since confirmed its capabilities. The US government, which was previously in a legal dispute with Anthropic, has also evaluated the model and now collaborates with the company on its deployment. The White House has established a process to test the most powerful AI models from leading firms before public release.
For most queries related to cybersecurity, biology, or chemistry, Fable 5 will route requests to the less capable Opus 4.8 model, which was made public in late May. Additionally, Anthropic has identified large-scale attempts to extract its technology for training competing AI models in authoritarian countries, and such queries will also be redirected to the lower-tier model.
Testing and Red-Teaming
The company hired outside experts to spend over 1,000 hours attempting to bypass these restrictions, a process known as red-teaming. It also ran a bug bounty program that paid individuals to find security flaws. According to Anthropic, no one successfully unlocked the model completely.
Pricing and Financial Context
Fable 5 comes with a significant price tag: $10 per million input tokens and $50 per million output tokens, double the cost of Opus 4.8. Tokens are the unit used to price AI model usage, and an intensive coding session can consume one million tokens in hours. Despite rapid revenue growth, Anthropic remains far from profitability and is paying a premium for computing power. It recently leased a datacenter from Elon Musk's xAI for $1.25 billion per month.
Industry and Political Context
Anthropic has been involved in a standoff with the Trump administration over its refusal to lift restrictions on mass surveillance and autonomous lethal weapons. Following this dispute, the Pentagon severed contracts with the company, whose AI tools had previously held defense security clearance. These developments occur amid intense financial excitement around AI. Both Anthropic and rival OpenAI recently filed IPO plans, while SpaceX is expected to break records with its market debut.
