Anthropic has released Fable 5, the first widely available model from its advanced Mythos lineup, while simultaneously restricting its use for sensitive cybersecurity and biological tasks. According to company reports, the San Francisco startup is routing high-risk queries to its lower-tier Opus 4.8 model to prevent the potential identification of vulnerabilities in critical infrastructure. The move follows months of internal testing and collaboration with the US government to mitigate security risks associated with high-capability artificial intelligence.
Why Anthropic is Restricting Access to Mythos 5
Anthropic has implemented strict usage boundaries for its Mythos technology due to concerns that the model could be used to exploit banking platforms or power grids. Anthropic officials stated that the model’s ability to analyze complex software code and identify security vulnerabilities poses a dual-use risk. To manage this, the company has deployed a “safety-first” routing system. Queries regarding sensitive topics like biology, chemistry, or cybersecurity are automatically redirected to Opus 4.8, which the company describes as a less capable, lower-risk alternative.
How the Cost of AI Models Impacts Enterprise Adoption
The release of Fable 5 introduces a premium pricing structure that reflects the high costs of training and deploying state-of-the-art AI. Anthropic has set the price at $10 per million input tokens and $50 per million output tokens. This represents a 100% increase over the cost of the Opus 4.8 model. Industry analysts note that while companies are eager to integrate these tools, the financial burden is significant. Anthropic is currently navigating its own financial challenges, including a reported $1.25 billion monthly lease for datacenter capacity from Elon Musk’s xAI.
The Growing Rift Between AI Developers and Government Regulators
The rollout of Fable 5 occurs against a backdrop of increasing friction between Anthropic and federal authorities. The Pentagon has severed its contracts with the company following a standoff regarding the firm’s refusal to lift restrictions on autonomous lethal weapons and mass surveillance tools. Despite this, Anthropic remains in alignment with the White House on other fronts; the company is collaborating with federal agencies to test its most powerful models for national security implications before they reach the public market.
Comparison: Current AI Model Tiers
| Model | Primary Use | Access Level |
|---|---|---|
| Mythos 5 | Advanced Research & Security | Restricted/Partner Only |
| Fable 5 | Public Coding & Analysis | General Public |
| Opus 4.8 | General Tasks/Safety Fallback | General Public |
Frequently Asked Questions
- What is the difference between Fable 5 and Mythos 5? Fable 5 is the public-facing version of the Mythos class, while Mythos 5 remains restricted to select partners, such as cybersecurity firms in the Project Glasswing program.
- Can users bypass the security restrictions on Fable 5? Anthropic reports that despite extensive red-teaming and a bug bounty program, no users have successfully unlocked the restricted capabilities of the model.
- Why is the Pentagon no longer working with Anthropic? The Pentagon ended its contracts after a dispute over Anthropic’s refusal to support autonomous lethal weapons and mass surveillance technology.
Stay Informed on AI Developments
Want to track how these models evolve in the coming months? Subscribe to our weekly tech newsletter for the latest updates on AI regulation, infrastructure costs, and industry IPOs.
