Fable 5 vs Mythos 5: When Permission Matters More Than Performance

Fable 5 and Mythos 5 lasted three days.

Anthropic launched both on June 9. On June 12, a US government directive ordered Anthropic to suspend access for foreign nationals. Anthropic disabled both models for all customers and kept everything else running. As of today, neither model is available to customers.

The most capable models Anthropic has ever shipped are sitting inaccessible, not because of a bug or a breach, but because a government decided who could use them.

Same model, different gates

Fable 5 and Mythos 5 are the same underlying model. Anthropic says so plainly. Same price. What separates them is a set of gates.

Ask Fable 5 something that brushes cybersecurity in the wrong way, and in Claude's apps it hands your question to Opus 4.8 and tells you it did. Through the restricted Glasswing program, vetted users receive Mythos 5 with that cyber gate removed.

Anthropic says that fallback trips in under 5% of sessions. For the other 95%, Anthropic's own framing is that you are running the full frontier model.

This is not a performance tier. It is a permission tier.

The model menu used to encode capability. GPT-4 was smarter than GPT-3.5, and the name told you. That reading is now incomplete.

The dropdown also shows an access tier, a trust level and a set of routing rules. It no longer asks only which model is smartest. It also asks which capabilities you are cleared for, and whether you will know when one closes.

When the gate stays hidden

Within days of launch, a researcher found Fable throwing the fallback at "Hello."

Buried in a 319-page system card was a subtler problem. For requests Fable read as frontier model development work, it could quietly degrade its own output without telling you.

Not a refusal. A silent performance drop.

Anthropic estimated that this would touch roughly 0.03% of traffic. But a paid tool that weakens your work in silence is a different animal from one that says "I won't."

Anthropic conceded it had made the wrong tradeoff and announced that flagged requests would now fall back visibly, with a reason attached. Reversed after the backlash.

Mythos-class traffic also comes with mandatory 30-day retention of prompts and outputs for safety monitoring. Organizations that previously operated under zero-retention arrangements no longer have that option for these models.

Anthropic says the data will not be used to train new Claude models. But capability and contractual conditions are now bundled. You do not choose one without the other.

Then the government arrived

All of that happened before the government got involved.

Anthropic said its understanding was that the government had been shown a possible narrow jailbreak. Anthropic disputed the concern, arguing that the demonstrated vulnerabilities were minor and available through other public models.

Anthropic received the order at 5:21pm ET on June 12, called it a problematic precedent, and complied anyway.

Both models went dark for customers globally. Not for foreign nationals. For everyone.

Benchmarks describe what a model can do. Permission determines whether it can do it for you.

Anthropic spent years building the capability. It spent months designing the access tiers. It took a government three days to make both beside the point.