Anthropic's model naming, extrapolated
Points and comments are a snapshot, not live.
Satirical extrapolation of Anthropic's literary naming scheme for Claude models as they grow larger and more complex.
The piece humorously extends Anthropic's pattern of naming Claude models after poetic forms (Haiku, Sonnet, Opus) into increasingly absurd literary and narrative references. Larger models are characterized with longer forms: Mythos becomes a scary Opus variant, Fable describes it until the question matters, Fable (xhigh) is labeled a "bankruptcy speedrun," and Saga adds meandering commentary. Further down the scale are increasingly meta constructs like Cinematic Universe with its Lore dispatch layer, Overwhelmingly Large Narrative Unit requiring a "previously on" segment, and Omnibus with continuous fine-tuning. The progression shifts from manageable literary structures to elaborate, self-aware fictional universes.
What commenters are saying
High-ranked commenters praise the naming scheme relative to OpenAI's alphanumeric approach. The thread branches into alternative name suggestions (Epic, Canto, Libretto, Corpus, Chant) and literary references. Commenters identify a Lovecraft allusion in Mythos and Iain M. Banks Culture series reference in "Overwhelmingly Large Narrative Unit." Discussion also surfaces criticism that Anthropic is degrading model capabilities rather than optimizing them, with some arguing newer competitors outperform Claude at lower cost. One commenter notes the original trio's initials map to behavior (Opus as OP, Sonnet as SO, Haiku as HA). Site design complaints about intrusive banners appear in a lower-ranked branch.