本地 Jev 复刻：7ms 决策跑贪吃蛇
开源 laya-mlx 在 Mac 上本地跑 Jev 式决策，单次 7–14ms、占用不到 1GB，贪吃蛇每秒 60 步，主打 agent 的 System 1 低延迟控制。
Open sourced "Jev " that runs 50x faster own Jarvis running locally now on mac 1GB RAM. 7ms per decision. laya-mlx : - 7-14ms decisions - under 1GB RAM - choice/score/yes-no, one forward pass, calibrated probabilities - no cloud, no tokens, no chat - 60+ moves/sec playing Snake on an M3 Max Now Why this hits? Agent loops die on latency and cost. If every “should I click, route, block, turn” takes 300ms in the cloud, you don’t get real-time control. 7ms local changes the architecture. This is System 1 for agents, Not another chatbot. A local brain that just picks the next action. - http://github.com/mizorewww/laya-mlx
这条是引用着另一条写的：「开源 Jev 替代 Laya：延迟与校准宣称超越 TypeSafe Jev」