What is Claude Fable 5? Anthropic's most powerful public model, explained

What is Claude Fable 5?

Fable 5, released June 9, 2026, is the first public model in Anthropic's most powerful class, the Mythos family. Until now that tier was locked behind a restricted partner program. Fable 5 is the same tech, opened up to anyone on a paid Claude plan, and it posts state-of-the-art scores on nearly every benchmark: software engineering, knowledge work, vision and scientific research. On some benchmarks Anthropic reports it scoring more than 10% above Claude Opus 4.8, the model the company shipped only weeks earlier. In plain terms, this is the most capable model a normal builder can put their hands on right now.

What is Mythos, and why did it spook governments?

Mythos is the model class underneath Fable 5. Anthropic introduced it in April 2026 through a restricted program called Project Glasswing, with partners including AWS, Microsoft, Apple and CrowdStrike. It rattled the cybersecurity world and national governments for one reason: it could autonomously find and chain zero-day exploits, the kind of offensive capability that used to require a skilled human team. That's why the full Mythos was never handed to the public. Fable 5 is the same capability tier, but wrapped in safeguards so it can ship. Anthropic also released Mythos 5, an upgraded Mythos reserved for trusted partners.

How do the safeguards work?

This is the whole reason Fable 5 can exist as a public product. In high-risk domains, specifically cybersecurity, biology, chemistry and distillation, Fable 5 doesn't answer with full Mythos power. It blocks the response and falls back to Claude Opus 4.8, a strong but less dangerous model. Anthropic says the safeguards trigger in fewer than 5% of sessions, so for ordinary coding, writing and analysis you're getting the top-tier model. The timing matters: Fable 5 landed just days after Anthropic publicly warned that AI is becoming too dangerous, which tells you the safeguard story is load-bearing, not marketing.

Is Claude Fable 5 free, and for how long?

For two weeks. Fable 5 is free in Pro, Max, Team and seat-based Enterprise plans through June 22, 2026. On June 23 it gets pulled from those plans and moves to usage credits, priced at roughly twice the current Claude Opus rate (about $10 per million input tokens and $50 per million output tokens). So there's a real, dated window here: the most powerful public model on the market is briefly free, then it isn't. Whatever you want to test against frontier capability, the cheap moment to do it is now.

The verdict: should you build on it?

Two things are true at once. First, use the window. Fable 5 free until June 22 is the cheapest your access to frontier capability will ever be, so take your hardest build, the one you weren't sure any model could carry, and pressure-test it on Fable 5 this week. If it works on the best public model and still falls apart, you've saved yourself months. If it suddenly works, you've found a real signal.

Second, don't confuse a better model for a business. Everyone gets the same Fable 5 at the same price. Your competitor can call the exact same API you do, which means the model is never your moat. The thing that's actually worth building is the non-model 80% around it: the proprietary data that compounds, the workflow you own end to end, the distribution you've earned, the trust that makes someone paste their real work in. That's the same test we run on every idea: if the model got twice as good tomorrow, does your product get more valuable or less? Fable 5 just made the model twice as good. Build the things where the answer is more.

So pressure-test on Fable 5 while it's free, then go own the part Fable 5 can't.

Related: Is it too late to start an AI startup in 2026? — a more powerful model doesn't close the door, it just closes the easy lane.

Frequently asked questions

What is Claude Fable 5?

Claude Fable 5, released June 9, 2026, is the first public access to Anthropic's most powerful model class, the Mythos family. It runs the same tier of capability as Mythos but is made safe for public use with safeguards that block responses in high-risk areas. It's state-of-the-art on nearly every benchmark, including software engineering, knowledge work, vision and scientific research.

Is Claude Fable 5 free?

Temporarily. Fable 5 is free in Pro, Max, Team and seat-based Enterprise plans through June 22, 2026. On June 23 it's pulled from those plans and requires usage credits, priced at roughly twice the current Claude Opus rate (about $10 per million input tokens and $50 per million output tokens). So the free window is two weeks, not forever.

What is the difference between Fable 5 and Mythos?

Mythos is Anthropic's most powerful model class, introduced in April 2026 through a restricted program called Project Glasswing with partners including AWS, Microsoft, Apple and CrowdStrike. It spooked the cybersecurity world for its ability to autonomously find and chain zero-day exploits. Fable 5 is the same capability tier wrapped in public safeguards. Anthropic also released Mythos 5, an upgraded Mythos for trusted partners.

Is Claude Fable 5 safe?

Anthropic's bet is that safeguards make Mythos-class power safe to ship publicly. In high-risk areas such as cybersecurity, biology, chemistry and distillation, Fable 5 blocks the response and falls back to Claude Opus 4.8. Anthropic says these safeguards trigger in fewer than 5% of sessions. The release landed days after Anthropic publicly warned that AI is becoming too dangerous, so the safety framing is doing real work.

Should I build on Claude Fable 5?

Use the free window to pressure-test your hardest idea on the most capable model a normal builder can touch. But don't mistake a better model for a business. Everyone gets the same Fable 5 at the same price, so the model is never the moat. Build the product where a smarter model makes you stronger, and own the non-model 80%: the data, the workflow, the distribution, the trust.

Source: Anthropic — Claude Fable 5 and Mythos 5 (June 9, 2026).