Anthropic Releases Claude Fable 5.2 and Mythos 5.2, Cuts Fable Input Price
Fable’s reported input price falls to $8 per million tokens. Anthropic also claims faster responses, while an optional faster-serving service costs extra.

AI-generated illustration by LightNews24.
Two new Claude models
Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 on November 9, 2026, with a lower reported input price for Fable. The company says both models improve coding, reasoning and vision—the ability to interpret images.
According to The Verge, Anthropic says both models generate outputs faster than their 5.1 versions. For Fable specifically, the company claims responses are 30% faster across effort levels than the previous version. Those are company performance claims, not independently verified results in the supplied reporting.
Fable’s lower input price comes with a separate speed option
TechCrunch reports that Anthropic is reducing Fable’s input-token price from $10 to $8 per million tokens. Tokens are the units used to measure material sent to a model. The change lowers the charge for that input, rather than establishing a lower price for every part of a request.
A separate faster-serving option is available for Fable 5.2 at double the price, TechCrunch reports. This concerns inference, the process of running a model to produce an answer. The report does not specify the full billing structure or how much additional speed that option provides.
The distinction matters for buyers: Fable’s claimed response-speed improvement and the paid faster-serving option are separate parts of the announcement. A lower input rate should not be read as a promise that every service configuration costs less.
Data-retention compatibility for enterprise users
TechCrunch also reports that Fable 5.2 supports zero data retention policies through Enterprise Frontier Safeguards. Such policies concern whether submitted data is kept, making compatibility relevant to organizations with strict data-handling requirements.
Compatibility does not establish that the policy applies automatically to every user. Eligibility and setup requirements are not specified in the supplied reporting, so organizations considering this feature will need those details before relying on it.
Share this story
See an error? Request a correction.