WHAT NEW IN X10.29 THU

X 上的 AI 合集档案 · 22 个合集 · 60,466 条

jevVIDEO

Jev 做本地 computer use:90ms 一次点击决策

用 CoreML 分割屏幕元素 + 本地 OCR,Jev 只吃文字标签就输出该点哪个按钮,全程不出本机、约 90ms 一次决策。

Milind S
@milindlabs2026.09.18收录原推发布于 2026.09.18
Milind S @milindlabs原推发布于 2026.09.18
在 X 上打开原推

Okay so Jev can actually do computer use really well Without any screenshots, or LLMs and no Pixels leave my mac I dont even read the Dom elements A local CoreML model segments every button and UI element on screen. On-device OCR reads the labels. That text is all Jev gets. It returns a probability across those elements and tells me the best one to click. Then it clicks, re-runs detection, and decides again. In a loop until the goal is done. ~90ms per decision. Faster than any LLM computer use I've tried. Blazing fast computer use, without any latency @typesafeai is building something really interesting

这条是引用着另一条写的:「ChatGPT 联合发明人发布新模型 Jev」

x.com/milindlabs/status/2100631847155994852采集于 2026.09.18 10:17