Anthropic's Model Naming, Extrapolated · Sam Wilkinson

With the release of Claude Fable, it is clear that Anthropic is progressing from poems to enterprise-scale narrative objects. To keep pace with competitors, the company is developing a broad portfolio of models optimized for the full literary stack.

Model Name<br>Description

Aphorism<br>One sentence, but it always feels right

Haiku<br>Small poem, small bill

Marginalia<br>Provides unprompted commentary on your code

Abstract<br>Summarizes reasoning it hasn’t done

Sonnet<br>Medium poem, medium bill

Diatribe<br>Sonnet, but angry

Opus<br>Long poem, entire bill

Treatise<br>Opus, but citation is left as an exercise for the reader

White Paper<br>Treatise, but you have to enter your email to get it

Mythos<br>Opus, but scary

Fable<br>Mythos, but the moral is about AI safety

Fable (xhigh)<br>Bankruptcy speedrun

Saga<br>Fable, but with extra meandering

Saga (Unabridged)<br>Includes answers to unrelated questions

Canon<br>Refuses to contradict prior answers

Lore<br>Interpretation requires a wiki

Cinematic Universe<br>Multiple Sagas with a Canon dispatch layer

Cinematic Universe (Director's Cut)<br>Same answer, 42% more tokens

Overwhelmingly Large Narrative Unit<br>Requires "previously on" segment prior to usage

Zach Snyder's Saga<br>Terminal turns black and white, becomes harder to follow