Ex-OpenAI researcher launches System One Models: Jev makes typed decisions in 70-500ms
JeremyCMorgan · x · 2026-09-25
TypeSafe AI, founded by ex-OpenAI researcher Diogo Almeida, unveils "System One Models" — a new class of models that skip text generation and output typed, probabilistic decisions software can consume directly. The first model, Jev, is in early access: it claims frontier-level intelligence on decision tasks at 70-500ms latency, two orders of magnitude faster and more efficient than LLMs, and hallucination-free since it only produces structured outputs. The stack includes a new architecture, a parallel sampler, and a training method called RLCD (Reinforcement Learning for Calibrated Decisions). The team discloses its eval caveats: reference answers are averages of two frontier LLMs and workflows were written in-house. Positioned as "a frontier-intelligence function call."
Related event: Ex-OpenAI researcher launches Jev, a non-text "System One" decision model(12 posts)→
More from Models
- Gemini 4 Carbon matches predecessor's score but explores less, gets distracted — patience_cave · 2026-11-11
- Fable 5.2 takes MazeBench frontier, matches GPT-6 Astra puzzle-solving — patience_cave · 2026-11-11
- MazeBench: Fable 5.2 scores 15%, near-frontier at half price; Gemini 4 Carbon 11% — patience_cave · 2026-11-11
- ERNIE 5.5 Flash Preview enters LMArena Battle Mode for crowdsourced ranking — arena · 2026-11-11
- Indie dev flips from $20 Pro back to $200 Gemini Ultra after two days of deep use — yihui_indie · 2026-11-11
- Leak: Google's upcoming model lineup includes Hydrogen, Krypton, Helix and a video model — koltregaskes · 2026-11-11