The post notes that Jev makes decisions rather than generating text—such as tool calls, retries, routing, or escalation—and argues that evaluating reliable autonomous decisions is a new challenge.
@typesafeai Jev makes decisions instead of generating text: which tool to call, whether to retry, route or escalate Interesting direction for agents. Because once software makes millions of these decisions autonomously, “valid output” ≠ reliable decision. That’s an eval problem!
Gregor Zunic showcases an open-source browser agent combining Browser Use and Jev, finding flights in 7 seconds for $0.0039, using a new action space per step, DOM state space, and small LLM fallback for typing.
Scott Williams predicts custom Jev-style models (parallel constrained decoding) will make existing company agent systems more token efficient, creating millionaires.
The user praises the JEV model for its extremely fast reproduction of each movement and command, believing bots will become smarter as a result.
