TypeSafe AI launches System One model Jev with RLCD for structured decisions

Why it matters

TypeSafe AI, founded by former OpenAI researcher and ChatGPT co-contributor Diogo Almeida, exited stealth with System One models and first public model Jev (early access). Officially: new RLCD training plus parallel sampling for typed probabilistic decisions software can consume directly—not free-form text—claiming near-frontier intelligence on System One tasks at ~20–200× speed and ~40–400× lower cost, with calibrated confidence and no type-error/hallucinated string generation by design.