把 Jev 放在每次 LLM 调用前做路由与分类

Laurent @Laurentxdev原推发布于 2026.09.19

在 X 上打开原推

A model that cannot write a single word, built only to decide Jev could sit before every LLM call for routing and classification Independent benchmarks will tell the real story Would you use it?

这条是引用着另一条写的：「ChatGPT 联合发明人发布新模型 Jev」