THE DIRECT ANSWER

On September 1, Anthropic introduced Claude Fable 5.1 and Claude Mythos 5.1. The company describes them as sharing an underlying model while applying different safeguards and access arrangements. Fable is generally available; Mythos access is restricted to trusted users for selected work, including areas with greater security implications.

Read the dated primary source ↓

Why the distinction matters

A model release is also an access-policy decision. For organisations choosing an assistant, the relevant questions extend beyond claimed capability to eligibility, permitted uses and the controls around deployment. A benchmark result does not explain those operating conditions.

What to examine

Read the terms for the specific release and workflow you intend to use. Treat the company’s performance claims as vendor-reported until independently assessed. The practical test is whether the available configuration improves a representative task while meeting the organisation’s requirements.

What this means in practice

The distinction between a shared underlying model and different access arrangements matters when interpreting capability claims. A result obtained under one safeguard configuration should not be assumed to describe another. Buyers should identify the exact product, access conditions and permitted use before connecting a release headline with their own deployment plans.

COMMON QUESTIONS

Questions readers ask

Does a shared model mean identical product access?

No. Products can share underlying research while differing in safeguards, eligibility and available features. Read the dated release's access conditions.

SOURCES & NOTES

Follow the evidence.

Source-based coverage of the dated announcement below, with Promptwire analysis. Company-reported claims are attributed to their source. The update date reflects added context and answers; it does not change the original event date.

  1. anthropic.com — official reference

Public sources support the factual background. Analysis and illustrative exercises are identified in the text. How we assess AI · Editorial standards

Spot a factual error or an outdated detail?