TypeSafe AI发布Jev：面向Agent循环的结构化决策模型
20 Agentic Use Cases of TypeSafe AI’s Jev
Last week, TypeSafe AI released Jev, its first System One model. Founder Diogo Almeida previously worked at OpenAI on the instruction-following research behind ChatGPT.
上周，TypeSafe AI 发布了 Jev，这是其首款 System One 模型。创始人 Diogo Almeida 此前曾在 OpenAI 工作，负责 ChatGPT 背后的指令遵循研究。
Jev does not chat, write code or summarize. It takes unstructured state and returns typed decisions with calibrated probabilities. That makes it a natural fit for the thousands of small judgments inside an agent loop: which model to call, whether a command is safe, which passage is relevant, whether the agent is actually done.
Jev 不聊天、不写代码或做摘要。它接收非结构化状态，并返回带有校准概率的类型化决策。这使其非常适合代理循环（agent loop）内部的数千项小型判断：调用哪个模型、命令是否安全、哪段内容相关、代理是否真正完成。
How Jev Works
Jev 的工作原理
Every call sends a state (text or JSON) plus a dictionary of typed questions. TypeSafe’s docs define 3 primitives:
每次调用都会发送一个状态（文本或 JSON）以及一个包含类型化问题的字典。TypeSafe 的文档定义了三种基本原语：
- Choice picks one option from a list and returns a probability per option plus confidence.
- Score rates the state on ordered rubric levels and returns probabilities plus confidence.
- Noul returns the probability (0 to 1) that a statement is true.
- Choice 从列表中选择一个选项，并返回每个选项的概率及置信度。
- Score 根据有序的量表等级对状态进行评分，并返回概率及置信度。
- Noul 返回陈述为真的概率（0 到 1）。
All questions are evaluated in parallel against the same state in one request. TypeSafe trains Jev with Reinforcement Learning for Calibrated Decisions (RLCD), so higher confidence should track higher accuracy. Choice supports up to 255 options.
所有问题均在一次请求中针对同一状态并行评估。TypeSafe 使用校准决策强化学习（RLCD）训练 Jev，因此更高的置信度应对应更高的准确性。Choice 最多支持 255 个选项。
The main claims, 193.6x faster and 444.6x cheaper, come from TypeSafe’s own workflow evals. The launch post says these figures sit on the higher end of real-world gains and use GPT-6 Astra and Fable 5.1 as the reference answer.
主要主张——速度快 193.6 倍且成本低 444.6 倍——来自 TypeSafe 自身的工作流评估。发布博文称，这些数字处于实际收益的高位，并以 GPT-6 Astra 和 Fable 5.1 作为参考基准。
Interactive Explainer
交互式解释器
更进一步：量化金融体系
看懂新闻只是起点——沿量化金融路径，把它变成能交付的工程能力