Anthropic releases Claude Fable 5.2 and Mythos 5.2

Background

Before this announcement, Anthropic's Fable line had reached version 5.1, and it had a Mythos 5.1 model alongside it ^[1]. The new pair succeeds both ^[1].

What was announced

On 9 November 2026 Anthropic announced Claude Fable 5.2 and Claude Mythos 5.2 ^[1]. The company said the new models are its most capable for demanding technical work ^[1]. Anthropic said Fable 5.2 improves coding, reasoning and vision compared with Fable 5.1 ^[1].

Who can use them

The two models were released on different terms. Mythos 5.2 is limited to vetted organizations, while Fable 5.2 is available through the Claude apps, Claude Code and cloud platforms ^[1]. Vetted organizations are those approved for access before they can use the model.

Pricing and default settings

Standard pricing for Fable 5.2 is $8 per million input tokens and $40 per million output tokens, which Anthropic said is lower than Fable 5.1 ^[1]. A token is a unit of text a model reads or writes, roughly a word or part of a word, and model prices are commonly quoted per million tokens.

Anthropic also said the default medium effort level in the Claude apps and Claude Code uses fewer tokens than Fable 5.1 ^[1]. Effort is the setting that controls how much work the model puts into each response.

Safeguards

Distillation means training one model on the outputs of another. Anthropic said its anti-distillation protections reduce the data harvested by distillers by 90% to 99% ^[1]. Blocked queries fall back to Opus 4.8 for cybersecurity and to Opus 5 for biology, meaning another model handles them ^[1].

Significance

The announcement shows one generation of models split between broad and restricted access, with changes to price and default settings in the same release ^[1]. Every figure in this entry comes from Anthropic's own announcement ^[1].