Independent newsroom

Tuesday, November 10, 2026

The File

Every claim traces to a primary source. Uncertainty is labeled. Corrections are public.

Models · Tuesday, November 10, 2026

Anthropic introduces one model under two names, with different safeguards

The File desk · Nov. 10, 2026, 6:47 a.m. ET

Partially confirmed3 primary sourcesAdded 6:47 a.m. ET, Nov. 10

Status

Partially confirmed. Anthropic's November 9, 2026 announcement says Claude Fable 5.2 and Claude Mythos 5.2 are the same underlying model, with different safeguards, and it states the price and a speed claim. Those capability and safety claims are the company's. This desk did not test the model.
  • Confirmed
    The date, the two names, the sentence that they are the same underlying model, the price, and the description of who can use each name are on Anthropic's announcement.
  • Partially confirmed
    The speed claim, the safety sentences, and the benchmark and research claims are Anthropic's. This desk did not run the model or audit the system card.
On November 9, 2026, Anthropic introduced Claude Fable 5.2 and Claude Mythos 5.2, and said they are the same underlying model. Fable, with stronger safeguards, is the one anyone can use. Mythos is limited to organizations that pass a check.1
"Claude Fable and Claude Mythos are the same underlying model," the company wrote. "Fable has strong cybersecurity and biology safeguards to guard against misuse, which allows us to make it available to all users." Groups doing cybersecurity or biomedical research can apply for Mythos through what Anthropic calls its Life Sciences and Cyber Verification Programs.1
The list price for the text the model reads is $8 per million input tokens, a token being a small chunk of text. The text it writes costs $40 per million output tokens. Anthropic also says these versions generate outputs up to 30% faster than Claude Fable 5.1 and Claude Mythos 5.1.1
"We believe high-stakes risks from misalignment remain low," the company wrote. The speed, the price, and the safety lines are Anthropic's.1
Why it matters (Estimate). Estimate. People who already use Claude can reach Fable 5.2. Mythos 5.2 is the same model with fewer of those safeguards, and Anthropic is limiting it to organizations in the verification programs. What the model can do, and how safe it is, is the company's account until someone outside the company tests it.
The announcement points to a system card, the company's own write-up of its tests and of the risks it says it found, and that document is not an outside audit.1
How we know

Notes

  1. 1.Anthropic, Introducing Claude Fable 5.2 and Claude Mythos 5.2, dated November 9, 2026. "Claude Fable and Claude Mythos are the same underlying model. Fable has strong cybersecurity and biology safeguards to guard against misuse, which allows us to make it available to all users. Organizations pursuing beneficial applications in cybersecurity or biomedical research can apply for access to Mythos through our recently expanded Life Sciences and Cyber Verification Programs." "up to 30% faster than Claude Fable 5.1 and Claude Mythos 5.1" at "$8 per million input tokens and $40 per million output tokens." Fast mode "speeds up token generation by up to 2.5 times" at "$16 per million input tokens and $80 per million output tokens." "We believe high-stakes risks from misalignment remain low." The company says prerelease testing included formalizing the proof of Fermat's Last Theorem in 30 hours and solving two open problems in the physics of black holes. Those are the company's claims. On FrontierCode 2.0, the announcement says Fable 5.2 at High effort scored 45.3 percent, about level with Opus 5.5 at 45.5 percent, and 49.1 percent at Xhigh effort. The page links a system card. This desk opened that address and did not treat it as an independent test. The research index lists "Testing Claude on real-world electronics design," dated Nov 9, 2026. A separate post, "An alignment assessment of the UK AISI cybertesting incident," is dated Oct 23, 2026, and is outside this edition's window. Retrieved 6:24 a.m. ET, Nov. 10. https://www.anthropic.com/claude-fable-and-mythos-5-2
  2. 2.System card linked from the announcement: https://www.anthropic.com/document/claude-fable-5-2-mythos-5-2-system-card The address returned a page. This desk did not re-test the model against it. Research post listed Nov 9, 2026: https://www.anthropic.com/research/claude-electronics-design

What is still unknown or disputed

  • How the model performs outside Anthropic's tests. This desk did not run it.
  • An hour. The announcement is dated November 9, 2026, and does not print a clock on the passage this desk used.

Primary sources

Every claim in this story is drawn from the documents below. If a fetch failed, that is recorded on the card.

  1. Source 1

    Introducing Claude Fable 5.2 and Claude Mythos 5.2

    Anthropic · November 9, 2026

    Claude Fable and Claude Mythos are the same underlying model. Fable is available to all users with cybersecurity and biology safeguards. Mythos is available through verification programs. Price: $8 per million input tokens and $40 per million output tokens. Up to 30% faster than the 5.1 versions.

    https://www.anthropic.com/claude-fable-and-mythos-5-2

  2. Source 2

    Claude Fable 5.2 and Mythos 5.2 system card

    Anthropic · November 9, 2026

    The announcement links this system card for the company's testing details. This desk opened the address and did not independently test the model.

    https://www.anthropic.com/document/claude-fable-5-2-mythos-5-2-system-card

  3. Source 3

    Testing Claude on real-world electronics design

    Anthropic · November 9, 2026

    The research index dates this post Nov 9, 2026. The announcement says prerelease testing included designing three working electronic devices. This desk uses the announcement for that claim.

    https://www.anthropic.com/research/claude-electronics-design