---
id: 20260917-T0-04
title: "Jev：只做决策分类路由的“系统一模型”，比小前沿LLM快100倍、便宜200倍"
title_en: "Jev: A System One Model 100x Faster and 200x Cheaper Than Small Frontier LLMs"
url: https://ai.daily.yangsir.net/daily/20260917-T0-04
issue_date: 2026-09-17
publish_date: 2026-09-16T11:09:53.000Z
category: news
source_name: "Latent Space"
source_url: https://www.latent.space/p/ainews-jev-a-system-one-model-that
---

# Jev：只做决策分类路由的“系统一模型”，比小前沿LLM快100倍、便宜200倍

TypeSafe推出Jev，一个被称为“系统一模型”的AI系统，专门负责决策、分类、路由和评分任务。与小型前沿LLM相比，Jev的速度快100倍以上，成本低200倍以上。该模型不追求通用生成能力，而是聚焦于高频、低延迟的判断类任务，适合需要快速路由和分发的场景。这一思路为AI系统架构提供了新方向：用专用小模型处理系统一任务，把大模型留给需要深度推理的系统二任务。

## English Version

**Jev: A System One Model 100x Faster and 200x Cheaper Than Small Frontier LLMs**

TypeSafe introduces Jev, a "System One Model" designed only for deciding, classifying, routing, and scoring. It is over 100x faster and 200x cheaper than small frontier LLMs. Rather than general generation, Jev targets high-frequency, low-latency judgment tasks suited for fast routing and dispatch. The approach suggests a new architecture direction: use specialized small models for System One tasks and reserve large models for deeper System Two reasoning.

---

**来源**：[Latent Space](https://www.latent.space/p/ainews-jev-a-system-one-model-that)

**详情页**：https://ai.daily.yangsir.net/daily/20260917-T0-04

---

*智语观潮 · Daily — https://ai.daily.yangsir.net/llms.txt*