Every tests a probability-output foundation model: 25x faster, ~600x cheaper than Fable
danshipper · x · 2026-09-16
Dan Shipper says Every has been testing a new kind of foundation model from typesafeai that outputs probabilities instead of words — like a code linter for knowledge work. In their tests it ran 25x faster at nearly 600x lower cost than Fable for similar jobs, usable as a judge for code quality, AI-isms in writing, or content interest. The quoted thread explores whether a writing model could use such probabilities mid-generation to mimic an author's word-by-word decisions, potentially a big upgrade for AI-assisted writing.
Related event: TypeSafe's Probability-Output Model Runs 25x Faster at 600x Lower Cost(4 posts)→
More from Models
- Periodic Labs pushes Kimi 2.5 base model past Astra with specialized scientific training — teortaxesTex · 2026-09-16
- OpenRouter spend flips to OpenAI over Anthropic for first time in 2.5 years — firstadopter · 2026-09-16
- KD in mid-training favors reasoning over factual recall, AI2/UW paper finds; Switch Distillation proposed — LukeZettlemoyer · 2026-09-16
- DoorDash, Siemens, Airbnb shift to cheap Chinese open-weight models — carlbfrey · 2026-09-16
- StepFun launches StepAudio 3: five audio models topping realtime voice leaderboards — StepFun_ai · 2026-09-16
- Abacus.AI says Smaug Flash fixes open-source models' tool-call hangs in production — bindureddy · 2026-09-16