Jev launches: a System One model for decisions
TypeSafe founder Diogo Almeida introduces Jev, its RLCD training method, and its focus on software decisions. The post includes a 2:56 launch video.

TypeSafe releases Jev, the first System One decision model that returns typed data (YES/NO, scores, choices) instead of sentences. It's fast: $0.042/M tokens and 70ms latency, filling the decision layer missing in AI agents.
On Sept 15, TypeSafe released Jev — the first System One Decision Model. It doesn't write sentences. It returns typed data: YES/NO, scores, choices. Fast. $0.042/M tokens. 70ms. This is the decision layer AI agents were missing.