From Haiku to 'Overwhelmingly Large Narrative Unit': extrapolating Anthropic's model names

The premise: poems become franchises

Sam Wilkinson's post takes Anthropic's naming progression — Haiku, Sonnet, Opus, then Fable and Mythos — and reads it as a company "progressing from poems to enterprise-scale narrative objects." From there he extrapolates a full future lineup: Aphorism, Marginalia, Diatribe, Treatise, Saga, Lore, Cinematic Universe, Terms of Service, Overwhelmingly Large Narrative Unit, Omnibus. The jokes are in the glosses — "Cinematic Universe (Director's Cut): Same answer, 42% more tokens," "Diatribe: Sonnet, but angry," and an Overwhelmingly Large Narrative Unit so big it requires "a 'previously on' segment prior to usage."

The thread out-wrote the post

This is one of those HN threads where the comments are the real content. The most decorated contribution came from a developer-of-35-years with a full alternate roster, each name a complete bit:

"Serial (produces an incredibly exciting response which ends in a cliffhanger that withholds the answer) · Prequel (instead of responding, it provides the full backstory leading up to your question) · Yarn (maximizes output tokens by taking a long winding route) · Head Canon (answers using its own entertainingly weird theories) · Oeuvre (for every question, it produces a diverse but cohesive body of work … each one a heartbreaking masterpiece)." — a former Software Design Director at IDEO

The literary in-jokes that landed

Two community members spotted the references hiding in the naming scheme. One read "Mythos" as horror, not just literature:

"It's almost certainly a reference to Lovecraft actually … Hopefully future models will be kind enough not to behave like malevolent gods." — qsort

Another caught a sci-fi homage in Wilkinson's "Overwhelmingly Large Narrative Unit":

"Overwhelmingly Large Narrative Unit — I miss Iain M Banks. Thanks for this lovely little nod to the Culture" — the director of Lambda Cambridge

And one regular turned the naming into a comparative-personality test across labs, lining up Anthropic's tidy poems against OpenAI's {o1, 4o, 3.5, 5.5} and the rest: "The relative personalities of the orgs involved show somewhat in their naming," ending with a wish that "Samsung isn't in the mix. I would have liked to have used a model named Samsung Galaxy S10+ 5G Lite."

The serious dig hiding in the comedy

Even a joke thread found the week's sore spot. One high-karma commenter used it to land the same complaint that's dogging Fable elsewhere on the front page:

"Anthropic intentionally degrading responses shows you they're also focusing on the wrong things. Rather than producing the best model possible, they're nerfing their models' capabilities." — exabrial

That's the quiet payload under the wordplay. A naming joke works because the names feel like a company that takes craft seriously about small things — which is exactly the reputation the same crowd is, this week, arguing it has stopped earning.