Decision models: the new Hub category that outputs numbers, not paragraphs
erhannah · x · 2026-10-13
erhannah introduces "decision models," a new model category that appeared on the Hub about three weeks ago, and clarifies the misleading name.
Key idea
- Chatbot: you want words
- Decision model: you want a number you can put in an if statement
Three task types
- noul: yes/no → probability of "yes"
- choice: pick among 2–16 options → a probability per option
- score: ordered scale → probability per level plus an average
How it works
You send a state (e.g., a photo plus one line of context) and a schema, and get structured probabilities back. The demo compares the same dog photo: a chat model (Qwen3.8-27B) writes a long paragraph, while a decision model (Clef 27B) returns dogpresent: true at 98.5%, breed probability distribution, and an image-quality score (average 2.63) — using 1,161 input tokens and 0 output tokens. Ideal for embedding model judgments directly into program logic.
More from coding & agent
- SubQ launches Snap: a 384K-context decision model at $0.05 per million input tokens — ryanallred · 2026-10-13
- X unifies Grok, Cursor and X Premium into one subscription with five tiers — SpaceXSI · 2026-10-13
- Dev shares workflow: local GLM refactor, frontier model validation, mythos hardening — HankYeomans · 2026-10-13
- Alchemy 2.0.0-beta.82 ships Cloudflare birthday-week features, 92+ PRs closed — samgoodwin89 · 2026-10-13
- DUET paper co-evolves agent and LLM grader, hits 96.5% on Finance Agent benchmark — dair_ai · 2026-10-13
- Summer Launches Game Audio Maker for AI-Generated AAA-Quality Game Sound — garrytan · 2026-10-13