LLMs

Claude Fable 5.2 vs Mythos 5.2: Same Model, One Locked

Anthropic shipped one model twice. The difference is what has been fenced off.

🕐 8 min read 👁 7 views 📅 Nov 22, 2026
The short version
● Same model Anthropic says Fable and Mythos are the same underlying model.
● The difference Safeguards. Fable has cyber and bio limits, so anyone can use it.
● Mythos Restricted to verified cybersecurity and life-sciences organisations.
● Price $8 in / $40 out per million. Fast mode doubles both.

This is the strangest pair Anthropic has shipped, because they are not two models. The company states directly that Claude Fable 5.2 and Claude Mythos 5.2 are the same underlying model. What differs is what has been fenced off.

Fable carries cyber and biology safeguards, which is what makes it safe to offer to everyone. Mythos is the same capability with those safeguards removed, and access is restricted to organisations that apply and are verified.

Side by side

Claude Fable 5.2Claude Mythos 5.2
Underlying modelSameSame
SafeguardsCyber and biology limits appliedRemoved
Who can use itAll usersVerified organisations only
How to get itGenerally availableCyber Verification Program or Life Sciences Verification Program
Enterprise routeStandardClaude Security, for Enterprise customers
Price$8 / $40 per 1M$8 / $40 per 1M
Fast mode$16 / $80 per 1M, up to 2.5x fasterSame
Speed vs 5.1Up to 30% fasterUp to 30% faster

Where Mythos actually scores higher

Anthropic publishes most benchmarks under Fable 5.2 and breaks out Mythos only where the safeguards bite:

BenchmarkMythos 5.2Mythos 5.1Opus 5.5 (no safeguards)
CyScenarioBench76.3%61.7%67.6%
CVE Rediscovery89.6%78.3%81.3%
Protein binder design, 24hr85.179.182.6
Terminal-Bench-Science 0.175.3%—58.7%

On CVE Rediscovery, Anthropic says Mythos 5.2 found 10.2% more vulnerabilities than Opus 5.5 and 14.4% more than Mythos 5.1. That is the gap the verification programme exists to control.

What Fable 5.2 does on everything else

BenchmarkFable 5.2Fable 5.1Opus 5.5GPT-6 Astra
FrontierSWE v269.9%56.3%62.3%65.5%
FrontierCode 2.049.1%32.4%45.5%42.8%
OSWorld 2.189.3%83.5%87.2%79.5%
Humanity's Last Exam, tools74.2%65.6%67.7%57.2%
Remote Labor Index41.3%17.9%21.3%20.8%
RoboDojo Sim64.9%27.6%39.4%29.8%

The Remote Labor Index jump is the one to look at — 17.9% to 41.3% in one version. RoboDojo Sim more than doubled. Those are not incremental.

Which one you can actually have

For almost everyone the answer is Fable, because Mythos is not purchasable. You apply, you are verified as doing cybersecurity or biomedical work, or you do not get it. There is no price difference to weigh, no tier to upgrade.

Anthropic's own guidance is to not start with either. It recommends Opus 5.5 as the daily driver and moving up to Fable only for the most ambitious open-ended technical work — and Fable can act as an advisor consulted by Haiku, Sonnet or Opus, or as an orchestrator delegating to smaller models, which is cheaper than running it on everything.

FAQ

What is the difference between Claude Fable 5.2 and Mythos 5.2?
Anthropic says they are the same underlying model. Fable has cyber and biology safeguards applied so it can be offered to all users; Mythos has them removed and is restricted to verified organisations.
How much does Claude Fable 5.2 cost?
$8 per million input tokens and $40 per million output. Fast mode is $16 and $80, for up to 2.5 times faster token generation.
Can I buy access to Mythos 5.2?
No. Access is through the Cyber Verification Program or the Life Sciences Verification Program, or via Claude Security for Enterprise customers.
Is Fable 5.2 faster than 5.1?
Anthropic says both 5.2 models generate outputs up to 30% faster than their 5.1 equivalents.
Should I use Fable 5.2 for everyday work?
Anthropic recommends starting with Opus 5.5 as the daily driver and moving up to Fable for the most ambitious open-ended technical work.
⚖ Our Verdict

For almost everyone it is Fable, because Mythos cannot be bought - only applied for. And Anthropic itself says to start with Opus 5.5 and move up to Fable only for the most ambitious open-ended technical work.