Chinese content开源模型 Jev 上线 7 天获 15K+ GitHub 星标
开源模型 Jev 上线才 7 天就拿了 15K+ GitHub 星，还冲上 Hugging Face 趋势榜第一，感兴趣可以去仓库看看。
#Jev
190 stories · 2 in 7 days · 54 in 30 days
Sep 26
Chinese contentJev 发布仅 7 天，开源 System 1 决策模型登上 Hugging Face 趋势榜第一
Jev 发布才 7 天，开源多语言 System 1 决策模型就冲上 Hugging Face 趋势榜第一，GitHub 星标破 1.5 万，追开源动态的可以看看。
LangChain webinar shows Jev enabling real-time inference use cases
Jev's speed and cost profile shifts evaluation and guardrail work from offline batch checks to inline, per-request processing. Teams building streaming or agent pipelines can apply the same pattern: filter content, enforce guardrails and steer agents during execution rather than after. The on-demand recording shows how LangChain's own team frames these trade-offs.
System One models reshape custom agent harnesses
The takeaway is that pairing a fast decision model with a reasoning model gives builders lower token costs and more predictable agent behaviour than either alone. Developers building agent harnesses can follow the linked guide to try Jev and CLM directly and compare their trade-offs in speed versus verification quality.
DSPy 3.4.0 adds native support for Jev System One models
System One models pairing with DSPy points toward faster, cheaper decision-making inside agent harnesses rather than just better text generation. Developers building routing, verification, or context-engineering layers can test the new integration in DSPy 3.4.0's compatible signatures. A hands-on resource is available at academy.dair.ai.
Chinese contentHarrison Chase：Jev 与 LangGraph 组合适合建模复杂智能体系统
LangChain 创始人 Harrison Chase 分享了 Jev 加 LangGraph 的搭配思路：用 LangGraph 搭系统骨架，Jev 处理小决策，搞智能体开发可以看看。
Chinese contentTypeSafe 发布首个 Decision 模型 Jev，驱动 Jev Router 智能分流
TypeSafe 做了个叫 Jev 的路由模型，每轮对话前给你的提示词打分，判断该用贵模型还是便宜模型，省钱和保质量二选一的问题它替你算。
Sep 25
Chinese contentChrome插件用Jev模型过滤X帖子，支持折叠、隐藏、置灰三种样式
一个Chrome小插件，加个过滤词就能自动折叠X上烦人的内容，免费能用500次，不喜欢马斯克刷屏的可以试试。
Chinese contentContrastive Language Model 发布：8B 参数，推理速度达 Jev 的 9 倍
做自定义 harness 的可以看看 CLM-8B，比 Jev 快 9 倍，DeepSWE 跑到 81.6%，代码和模型都开源了。
Chinese contentLatent.Space 对谈 Jev 创作者：为何不做聊天优先的 AI
Latent.Space 访了 Jev 的创造者，聊这个专门在软件里做可靠决策的模型为什么故意不走聊天路线，还吐槽公开基准，观点挺反主流的。
Sep 24
Douchat adds Jev to coordinate multiple group-chat agents
Coordination here is split cleanly: structured judgement goes to Jev, planning and tool use stay with each agent's own model, and a program layer owns dispatch, timeouts and persistence. That pattern is reusable for anyone wiring multi-agent systems into chat platforms like Telegram or WeChat. The fallback design, where ambiguous or failed decisions degrade to the members' LLM, is a practical detail worth borrowing.
Chinese content实测：3分钟 AI 宣发视频可被 oneshot 完整克隆
有人把 Jev 的 3 分钟宣发视频 oneshot 复刻出来了，人物声音动作都能改，看完你就懂爆款视频现在多容易被复制。
Jev-as-a-Judge for agent evaluation, with frontier model escalation
The tiered judging setup suggests evaluation cost can be cut without giving up accuracy, since most verdicts likely land in the high-confidence tier where the cheaper model suffices. Teams building agent pipelines can copy the escalate-on-low-confidence pattern and apply it to their own eval harnesses. Comparisons against GPT-6 and Opus 5.5 as escalation targets give a concrete baseline for benchmarking.
Chinese contentJev 用作放射报告事实性评估器：RadEvalX 相关性达 0.573
医院里 AI 写放射报告容易漏病灶或编造发现，这篇论文用 Jev 做低成本裁判，成本不到 3 美分评 100 对报告，做医疗 AI 评估的可以看看。
Sep 23
Chinese contentTypeSafe AI 发布判别式模型 Jev：每百万输入 token 仅 0.042 美元
一个只会做选择题的模型，100 万次判断才 42 美元，Agent 里几十个判断步骤全靠它省钱省时间，Vercel 上线一天就 13% 付费团队接入了。
Chinese content研究：类型安全决策模型仍会受选项名称语义干扰
做结构化输出的人该看看：选项叫 0/1 还是 no/yes 能让 AUC 从 .94 掉到 .23，类型校验根本拦不住这种错。
Jev tested across three Zilliz open-source RAG projects
The benchmarks show where cheap small models can replace larger ones in RAG pipelines: stopping decisions and relationship filtering see large latency and cost savings with matched recall. Engineers building retrieval systems can use the published code to pick where to route between Jev, GPT-5-mini and Voyage rerank-3. The evaluation includes Chinese-language queries, useful for teams serving bilingual users.
llm-typesafe 0.1a0 plugin adds Jev model to LLM
The interesting part is the answer-type interface: Jev is accessed through structured noul, choice and score outputs rather than free-form chat, which suits classification and triage pipelines. Developers already using Willison's LLM CLI can wire it in with two commands and pipe files straight into scoring workflows. Compared with general chat models, this targets routing and labelling tasks where a typed confidence value is the whole output.
Chinese contentJev是什么：TypeSafe AI 推出只做判断不输出文本的模型
前 OpenAI 研究员搞了个不会打字的模型 Jev，专门做分类和判断，几十毫秒出结果还带置信度，比拿 ChatGPT 硬干省快多了。
Chinese content用 Jev Classifiers 自动分类 LLM 请求
OpenRouter 介绍了 Jev 的 Classifiers 功能，能自动给你的 LLM 请求打标签，再到 Explore 页面看各场景的用量分布，运维排查很方便。
Sep 22
Papers12:34
Chinese contentJev 科学决策评测：12 个配置对比语义选择，延迟最低有人把 Jev 当科学决策组件实测了一遍：20 道语义题跟 12 个配置比，它全对且延迟最低。
Chinese content用 Jev 决策模型替代大语言模型做边缘服务编排，延迟降15.9%-26.5%
有人拿 Jev 决策模型替掉 DeepSeek 和 Qwen 跑边缘编排，延迟最多降26.5%，API 费用省近七成。
Chinese contentHugging Face 创始人：90% 真实场景里 LLM API 并非最优选择
Hugging Face CEO 说九成场景用 LLM API 太浪费又贵又慢，Will Depue 还提到 zero-shot 分类器 Jev 这个老 ML 回潮的路子。
TypeSafe launches System One model Jev
Jev offers a different approach to AI decision-making that could better handle everyday tasks, contrasting with text-based models like Claude Code and Cursor.
Jev vs Reranker in RAG
Jev shows promise for offline retrieval and data cleaning but struggles with latency-sensitive online RAG due to high API overhead.
Chinese contentLovable 上线 Jev API 与 Playground，本周免费用
Lovable 送 Jev API，本周建 agent 应用就能领，Playground 可先试。
Chinese contentTypesafe AI 的 Jev 模型上线 OpenRouter
主打快和便宜的 Jev 模型上了 OpenRouter，适合塞进工作流里，专干那些要快、要省钱的判断活。
TypeSafe AI launches Jev decision model
Jev offers a cheaper alternative for classification tasks like spam detection and search reranking, with output tokens free and input pricing competitive.
Chinese contentJev 在 Vercel AI Gateway 免费开放至 9 月 25 日
Jev 在 Vercel AI Gateway 免费到 9 月 25 日，官方还配了表单路由模板，想自动分流表单提交可以直接套用。
Chinese contentTypeSafe 详解 Jev：RLCD 训练的可靠决策模型，告别聊天优先
TypeSafe CEO 在播客里讲为什么 Jev 不做聊天、专攻软件内的可靠决策，连公开基准都拒绝跑，观点挺反共识。
Jev research paper discovery tool
Jev offers researchers a practical way to navigate academic papers through agent-assisted tagging, potentially saving hours of literature review time.
Jev matches human reviewers in agent evaluation
Jev offers 98% cost savings and 5x faster evaluation compared to Claude Sonnet 4.6 for developers testing AI agents.
LangChain adds Jev middleware
Developers in ASEAN can now simplify model routing for their LangChain agents without manual selection.
Jev organizes 2.3K AI research papers
Jev offers a cost-effective alternative to expensive LLM classifiers for organizing research papers, costing only $0.14 for 2,300 papers.
Jev Model Explained For Agents
Jev offers a straightforward way to enhance agent capabilities; developers can implement it quickly following this guide.
Sep 21
TypeSafe AI Opens Jev Access With No Waitlist
Sign-up no longer requires a waitlist, so anyone can create an account at console.typesafe.ai and try Jev the same day.
