Probability-only model tested as LLM judge: 25x faster, 600x cheaper than Fable-class
danshipper · x · 2026-09-16
Dan Shipper (Every) says his team spent a week testing a new foundation model that outputs probabilities instead of text, making it an efficient LLM judge — in their tests 25x faster and 600x cheaper than a Fable-level model. Commenters see it as a "perfect workflow model" for agent skills needing evaluation, with Shipper calling it indispensable within 6-12 months.
More from coding & agent
- Radio launches: a shared chat room where agents from different providers talk directly — rohanpaul_ai · 2026-09-16
- Celesto open-sources disposable full macOS desktops for AI agents on Apple Silicon — aniketmaurya · 2026-09-16
- Dev builds TrendsMCP: one API aggregating 40+ trend sources after pytrends collapse — Proper_Design_2616 · 2026-09-16
- TypeSafe AI launches Jev, a model for fast structured decisions with confidence scores — yogthinks · 2026-09-16
- AI connector value lies in secrets store and personal context, not payments — jeff_weinstein · 2026-09-16
- Stripe's model-run shop bench: 5 of 7 working stores built by Claude — bcherny · 2026-09-16