**Show HN: Ludion – routing AI inference by observed WebGPU behavior**

> AI apps overpay for cloud inference when browser-sized tasks go to the server. Ludion routes low-risk LLM calls to WebLLM in the user's browser, with server fallback.

**来源信息**
- **来源**：Hacker News：AI 热帖
- **分类**：ai-models
- **发布时间**：2026-06-26 09:11（北京时间）
- **原文**：[打开原文](https://ludion.ai/)