TypeSafe AI Launches Jev, a Machine-Native Model for Typed Probabilistic Software Decisions
Summary
TypeSafe AI launches Jev, a machine-native model built to return typed probabilistic software decisions rather than text, with CEO Diogo Almeida saying its calibrated-decision architecture can act on structured data such as Doom game states in 70–500ms.
Key Points
- TypeSafe AI launches Jev, a machine-native model that returns typed probabilistic decisions for software and can play Doom from structured game-state data.
- Co-founder and CEO Diogo Almeida, a former OpenAI researcher and RLHF co-inventor, says Jev uses a Reinforcement Learning for Calibrated Decisions architecture rather than generating natural-language text.
- TypeSafe claims Jev responds in 70-500ms and charges $0.042 per million input tokens with no output-token fee, versus OpenAI GPT-5.6 Terra's listed $2 input and $12 output rates.