The author spent a day testing whether JEV could route AI agents better than major models. Initial results lost, but after changing the question format it tied them, at one-third the latency and a fraction of the cost. Includes full measurements and where it fails.
I spent a day trying to prove JEV could route my AI agents better than the big ones. It lost. Then I changed how I asked the question and it tied them, at a third of the latency and a fraction of the cost. Everything I measured, including where it fails 🧵
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.
