XGames and real time
Updated Nov 13, 2026
What people are building with Jev
855 projects, posts and guides about Jev, the System One model from TypeSafe AI. Each entry links to its source and shows the cost and speed its author reported.
- 855builds
- 182guides
- 21use cases

Diogo Almeida
@CompleteSkeptic
After co-inventing ChatGPT, I kept asking myself: why have superhuman chat models not led to AGI? I’ve spent the last 2 years in stealth building a new way to train models (RLCD), and a new type of frontier AI model that we are releasing today: Jev • 20-200x faster • 40-400x cheaper (w/ output tokens free) • Frontier composable intelligence optimized for decisions AFAICT the shortest path to AI-based economic revolution
Open the post on X 
Frit🅾️ Pendej🅾️
@vwapster
I think Jev and Laya were the missing pieces to algorithmic trading. Sure you can hardcode gates and adjust variables, but being able to add a real time decision maker within your algorith is a game changer. I can smell and taste an early retirement.
XTrading and markets
Real-time decisions in algorithmic trading

Frit🅾️ Pendej🅾️
@vwapster
Jev usage on 1 trading bot. It's cheap but this is going to add up fast. @deepseek_ai please help and release something similar for free.
XTrading and markets
Cheap per call, but it adds up

Jason Zhu
@GoSailGlobal
拿 Jev 做搜索重排,我先泼一盆冷水:单独用,它没打赢向量检索 TypeSafe 的 Jev 这阵子很火,一堆项目拿它做重排。我们在 Agent Skills Hub 的 33,047 条目录上认真测了一次,164 条中英文真实查询,9,831 对分级标注,整套只花了 2.6 美元 三个结论 01|单独重排,约等于没赢 Jev 重排 bge-m3 的前 30 条,NDCG@10 只多了 0.012,置信区间跨过零。MRR 和前三命中率倒是涨得明显,它很会把最强的那一条顶到第一,后面几条基本是重新洗牌 02|裁判偏差,被我们量出来了 Jev 自己也参与了打标注,这就是循环 只用 Jev 当裁判,它领先 0.053 两个裁判合并,领先 0.012 只用 Haiku 当裁判,反而落后 0.028 同一组比较,换个裁判结论直接翻面。所有涉及 Jev 的结论,我们只认 Haiku 那一列 03|真正稳赢的是融合 把 Jev 和 bge-m3 的排序做 RRF 融合,NDCG@10 到 0.864,比纯向量高 0.064 到 0.116,三种裁判下都成立。代价是每次查询多一次 API 调用,约 0.0002 美元 顺便暴露了我们自己的问题:Hub 线上 CLI 用的关键词排序只有 0.609,短板是召回。相关结果有一半压根没进候选,后面怎么重排都救不回来 数据、标注、每条查询的得分全部开源,不用 API key 就能复现打分
Guidex.com
Jev as a reranker: an honest negative
Jason Zhu tested Jev reranking on 164 real queries. Alone it did not clearly beat vector search; fused with it, it did. In Chinese.

Frit🅾️ Pendej🅾️
@vwapster
Grok bot + Jev + Laya. Self improving and auto calibrating Kalshi Solana trading bot. Still on paper. Looks promising.
XTrading and markets
A Kalshi and Solana bot on Grok, Jev and Laya

Vintuxai
@vintuxai
Stop settling for mediocre AI outputs. 🤖 Introducing slop-grader: The new Jev-AI CLI tool designed to audit text against your specific rulesets. It’s time to take control of the "slop." 🛠️ Here is the breakdown 🧵 #AITools #ProductivityAI
XSocial feeds
slop-grader

恒星sun
@vintcessun
如果分类结果最终要喂给业务系统,生成一段 JSON 往往不是最省事的路径。Simple Jev 把开放模型变成结构化决策接口。 https://github.com/featherless-ai/simple-jev 它复用共享上下文的 KV cache,只读取各问题候选标签的下一 token logits,再由服务端组装 choice、score 和真值判断。省掉逐 token 解码与 JSON 解析,多问题场景更有价值;common 目录还把提示契约和评分逻辑抽成可复用层。适合路由、审核、打分,但准确率与校准仍需按任务验证。
XOpen source
Simple Jev, reusing the KV cache

Jiayuan (JY) Zhang
@jiayuan_jy
一些关于 Jev (@typesafeai) 的 notes 研究了一天 Jev,带来的新鲜感迅速回落,这好像就是一个更快的通用分类器/决策器,LLM 完全可以做到。 而且因为不了解模型背后的参数规模(应该不会很大),世界知识可能不一定有常规 LLM 那么全,在这种情况下,它的复杂场景的决策结果是不是真的准还是要打一个问号的。 在一些有限解空间 + 低延迟要求的问题上,Jev 可能是一个很好的方向,加上形式化输出从程序上保证了正确性。 一个最常见的场景就是 Computer Use,太适合 Jev 了,网页的 Dom 元素是一个有限集,完全可以让 Jev 来做操作,这里面的 loop 相当于是 dom list -> jev action -> new dom list 这样的循环,但是这里 Jev 的推理能力和长上下文情况下的 computer use 效果如何还不清楚,目前还没有看到 benchmark(直观判断肯定是不如 GPT 6 Astra 的,但是速度太快了)。 昨天有尝试把 Pi Agent 中间的一些模块用 Jev 来重写一下,发现可以优化的地方不是特别多,tool using 部分的选择还是不能用 Jev 来代替,因为这不是一个有限集(每一步 tool using 实际上带了很多参数,比如 edit tool,会有具体的 lines 等参数,这部分是需要模型推理出来的,没有办法提前加到 Jev 的决策列表中),但是有一些地方是可以的,比如说 Compaction,可以让 Jev 快速做分类器(LLM 也能做,这里面差异化不太大)。 另外一类场景是依赖决策树逻辑的,例如: - 游戏 AI - 机器人 - 自动驾驶 而且这些场景对实时性要求比较高,传统的 LLM 来做这些事情的一个很大问题就是太慢了(plus 很大一部分场景是缺乏数据来训练的)。 目前正在做的两个 demo: 1. Poker AI,Poker 非常适合这个场景,而且决策树非常长 + 复杂,正在用 Jev 和其他模型做对抗式 battle。(btw 传统的 GTO Wizard 用来做训练实在是太难用了) 2. Pokemon VGC AI,这是严肃的宝可梦双打对战,有世界锦标赛的那种,每个赛季都有对应的规则,因为数据很全,所以非常适合用来研究 AI 的决策,plus 之前竟然没有一个很好的用来个人训练的 AI(这个做完了打算用这个 AI 实际在 Pokemon Champion 排位赛里用一下)。 这两个 demo 场景都是偏回合制的,其实用 LLM 也能做。
Guidex.com
Day-one notes from a skeptic, in Chinese
Jiayuan Zhang: a faster classifier that an LLM can also be, but a good fit for computer use, games and robotics where the options are known.

VertrAI
@vertr_ai
The useful engineering lesson is the async split: planning can continue while Jev selects the next action. That makes sense without assuming it proves AGI. We explain the roles using this demo in our 105-second video: x.com/vertr_ai/statu…
XAgents and browsers
Planning while Jev picks the next action

VertrAI
@vertr_ai
The small-LLM fallback for typing is an important detail: Jev chooses actions, while text generation stays with an LLM. We featured your 1x demo, with credit, in our short explainer of that division of labor: x.com/vertr_ai/statu…
XBenchmarks and evals
Jev chooses, the LLM writes

VertrAI
@vertr_ai
154ms vs 860ms is a useful routing result, scoped to these 200 synthetic classification cases—not a blanket ChatGPT comparison. We credited your latency and cost charts in our Jev video, alongside the game/browser demos: x.com/vertr_ai/statu…
XBenchmarks and evals
154 ms against 860 ms, and the scope of it
- Routing
- 154 ms vs 860 ms
- Cases
- 200 synthetic

Jason Zhu
@GoSailGlobal
一个只会做选择题的小模型,能长出多少玩法? TypeSafe 的 Jev 就是这么个东西:给它最多 255 个选项,几百毫秒挑一个,再附一个校准过的置信度,输出 token 免费 结果开发者们已经拿它做出了 19 个开源项目,加起来 6800 多星 挑几个最有意思的 jev-ultrafast,browser-use 出品,2700 星,一句 slogan:i. am. speed. openjev,想在家里一张 3090 上跑出类似 Jev 的东西,925 星 fast-jev-compaction,替换 Claude Code 的上下文压缩,每次工具调用逐条打分,过期的丢,留下的原文保留 typesafe-computer-use,操控 Mac 每一步只要 0.0002 美元 jev-router,在 Claude Code 里按任务自动路由到最便宜的模型 pg-jev,Postgres 扩展,用大白话查表 还有人拿它打超级马里奥、在 MuJoCo 里以 2.5Hz 开无人机 我们在 Agent Skills Hub 把这 19 个放进了一张对比页,星数、语言、协议、质量分一张表看完,每 8 小时按 GitHub 实时数据重算 两个提醒: 1️⃣ 19 个都过了 Hub 的基础规则扫描,但那是规则扫描不是人工审计,装之前还是看一眼代码 2️⃣ 另外有 4 个项目没写开源协议,拿去商用前先确认
Guidex.com
19 open-source Jev projects
Jason Zhu’s roundup of 19 open-source Jev projects with about 6,800 stars between them. In Chinese.

VertrAI
@vertr_ai
ChatGPT thinks through the plan. Jev helps pick the next move. ⚡ Minecraft, browser automation, speed and cost—Jev explained in under 2 minutes. 👇 Featuring demos by @rronak_ and @gregpr07, benchmarks from @OpenRouter, and Jev by @typesafeai. #Jev #AI
XGames and real time
The next move in Minecraft and in a browser

Vadim Choi
@vadimchoi
$10k/month just to decide which agent should answer. not to answer. to decide. group chat + @mentions: fast, cheap, terrible UX. an LLM orchestrator: nice UX, 4-7s and $0.00046 a message. @typesafeai jev: 0.29s median, $0.00002. same decision, $20/month.

Utkarsh Upadhyay ⚒️
@utk2103
just shipped jev-studio v0.2.0 @typesafeai Jev in one pip install - MCP tools for Choice / Noul / Score - `jev` CLI now with dry-run provenance - ready-made prompt libraries + slash commands for every cookbook - Claude Code + Codex plugin manifests pip install jev-studio
XSDKs and integrations
jev-studio v0.2.0

Moritz Kremb
@moritzkremb
All of the coolest Jev projects I could find on X today 🧵
Guidex.com
The coolest Jev projects on X
Moritz Kremb’s thread of the best Jev projects he found on X in one day.

Ayush Jain
@urayushjain
Most AI models are built to generate. But Jev is trying something different: Don’t generate. Decide. Give it some context and a specific question, and it returns a structured decision with a probability. I found this interesting for things like model routing, agent workflows, tool safety and RAG. The idea is simple --Instead of calling a big LLM for every small decision, let your code handle the workflow and use a smaller model where human-like judgment is actually needed. Honestly, this feels like an interesting direction for AI systems.
XRouting and model choice
A structured decision with a probability

梅澤 寛太|Web・業務システム開発
@umezawakanta13
Jev×マリオ風1-1、検証途中の録画です。最初の敵に接触してミス。回避補助の導入前の映像です。 別試行はブラウザ往復中央値976.8ms(5応答)。純粋な推論時間ではありません。 検証記事:zenn.dev/kanta13jp1/art… 体験:my-web-app-b67f4.web.app/jev-mario-lab

梅澤 寛太|Web・業務システム開発
@umezawakanta13
計測画面付きの別試行も補足します(成功3応答)。 中央値: ・ブラウザ往復 1,139.7ms ・Jev HTTP往復 490.3ms ・状態取得→操作 1,139.8ms 今回も最初の敵に接触。HTTP往復には通信時間が含まれ、純粋な推論速度ではありません。遅延と操作判断の影響は引き続き切り分け中です。
XGames and real time
Where the latency actually goes
- Browser round trip
- median 1,139.7 ms
- Jev HTTP
- 490.3 ms

小墨同学
@xiaomovps
Jev 发布没几天,开源社区已经开始疯狂复刻了🔥 最值得推荐的五个模型: 1、Laya 421M:原生决策模型,支持 Mac 2、Decider-2B:最像 Jev,基于 Qwen3.5 3、NanoJev 0.6B:专门的 Decision Head 4、Reflex:Qwen3.5 + Direct Logits 5、System-One 4B:专门做概率校准 如果和我一样是苹果的芯片,我推荐: Laya 和 Reflex 下一步我准备选两个在本地运行 然后测试一下和Jev的差距
Guidex.com
Five open Jev replicas, in Chinese
小墨同学 on Laya 421M, Decider-2B, NanoJev 0.6B, Reflex and System-One 4B, and which run best on Apple silicon.

uehaj
@uehaj
jeven-doors.uehaj.workers.dev/?g=mm2rc2cp 「Jevの扉」、ジャンルを作成してアップロードする機能をつけました。WebMCP対応なので、エージェントに作成させると便利です。 #Jev #JevenDoors
XTools and apps
Jev Doors takes uploaded genres

つぼね👨💻
@tubone24
Jevに飽きちゃったのと、全然利用料が安いので、公開してみました。 雨の日はスピードで遊ぼう!! jev-speed.tubone24.workers.dev
XBenchmarks and evals
A Jev speed test app, made public

tshmieldev
@tshmieldev
Jev vs LLMS for clasifying tweets in your feed! Try it out: github.com/tshmieldev/sha… Make sure to drop a star!
XSocial feeds
Jev against an LLM on your own feed

思维怪怪
@0xLogicrw
Jev 这个 waitlist 还是很给力的,昨天申请,今天就能用上。 给已经拿到 API、但还不知道怎么玩的人整理了一份 Awesome Jev,目前我能确认到的 Jev 项目基本都在这里: 1. jev-ultrafast Browser Use 做的高速浏览器 Agent。Jev 负责每一步判断「做什么、点哪个元素」,需要输入文字时才调用小模型。Google Flights 搜航班完整跑完约 7.1 秒。 https://github.com/browser-use/jev-ultrafast 2. typesafe-mcp 把 Jev 直接接进 Claude Code、Claude Desktop 和 Codex。Agent 可以随时调用 Jev 做 Choice / Score / Noul 这类结构化判断。 https://github.com/itsmostafa/typesafe-mcp 3. jev-mcp 另一个更偏实用工具的 MCP,已经封装好事实核验、Prompt Injection 检测和语义排序。 https://github.com/jkudish/jev-mcp 4. SemDecide 把 Jev 做成 Unix 命令行工具,可以直接拿来做语义判断、分类、打分、过滤和 Guard。很适合塞进脚本、CI 和数据流水线。 https://github.com/sharziki/semdecide 5. Jev Codex Router 先让 Jev 判断每一轮编程任务有多难,再自动决定交给便宜模型还是更强模型。作者用 237 个真实 turn 回放,自测成本降低约 60%。 https://github.com/0xNatoshi/jev-codex-router 6. Winnow 给 Claude Code 做「上下文垃圾回收」。Read、Bash、Grep 输出太长时,先让 Jev 判断哪些内容真的跟当前任务有关,再决定哪些需要进上下文。 https://github.com/GhalebDweikat/winnow 7. Jev Review 用 Jev 做代码审查前置筛选,先判断 correctness、security、reliability、兼容性和测试风险,再把重点问题交给更重的模型处理。 https://github.com/devagrawal09/jev-review 8. Blink 用 Jev 在代码库里做语义寻路。每到一层目录,就判断哪些文件或文件夹最可能和问题有关,再把更多 walker 分配过去。 https://github.com/ellipsis-dev/blink 9. neo4jev 把同样的思路用到知识图谱。Jev 给每条候选边打概率,再用 beam search 一步步找到最可能的路径。 https://github.com/jexp/neo4jev 10. jev-desktop 把 Jev 接到桌面自动化里。它读取 Accessibility Tree,判断该操作哪个控件、执行什么动作,再交给本地执行器。 https://github.com/lahfir/agent-desktop 11. TypeSafe AI Playground 一个专门拿 Jev 做实验的 Rust CLI,已经做了 PHI 检测、代码注释审查、实时语气分析、职业和行业分类等玩法。 https://github.com/markjaquith/typesafe-ai-playground 12. Prism 目前我看到最接近量化/交易场景的项目。它让 Jev 判断流动性池的 toxic flow、市场压力、均值回归可能性和流动性分布方式,目前默认还是 shadow/advisory,不直接驱动交易。 https://github.com/irfndi/prism-liquidity-agent 13. 1v1 Jev 直接做了一个 FPS,让 Jev 大约 9Hz 判断移动、瞄准、ADS、开枪和跳跃。属于最直观展示 Jev 低延迟决策能力的项目之一。 https://github.com/emrickgarrett/OneVOneJev 14. TypeSafe on Neon 拿 Jev 做模型 Router。请求先由 Jev 判断属于哪种任务,再转给 Grok 4.6 或 GPT-6 Astra 等模型执行。 https://github.com/andrelandgraf/typesafe-on-neon 看完这些项目,你大概也知道 Jev 可以怎么玩了,就是拿它反复做「要不要、选哪个、打几分、下一步干什么」。 真正需要生成代码、写文章和复杂推理时,再把任务交给传统大模型。
Guidex.com
Awesome Jev, in Chinese
思维怪怪’s hand-checked list of early Jev projects, with a walkthrough of each.

Shashank Jha
@theshajha
if you think you have seen it all with jev (@typesafeai), here's something new - play chess against jev
XGames and real time
Play chess against Jev

Akshay 🚀
@akshay_pachaar
A faster, 100% open-source alternative to Jev. It’s called Laya. Both models solve the same problem. You give them unstructured state plus typed questions. Instead of generating a paragraph, they choose from answers you define and return probabilities your code can act on. But they make those decisions differently. Jev is a closed, hosted API. Its architecture and training are not public, but it works zero-shot and supports up to 255 options. Laya is Apache 2.0 licensed, runs locally, and uses open-weight encoder models. Here is how its pipeline works. → It routes by language first. Plain Python selects the English or multilingual checkpoint before inference. → It creates one sequence for each question. Every possible answer gets a `[MASK]` marker. The question and state provide the context needed to compare those options. → It scores instead of generating. A bidirectional encoder and small decision head assign one score to each option. Softmax turns those scores into a probability distribution. → It batches the questions. They run together in one model call, although every question still carries its own copy of the state. This removes network latency and token-by-token generation. In one independent test, Laya took about 35 milliseconds on a local GPU. Jev took roughly 380 milliseconds over the network. Laya also has no per-call API fee, keeps data on your infrastructure, and can run on older GPUs. But it is not a drop-in replacement for every workload. Jev is a hosted generalist that works well out of the box. Laya is a fast base you need to specialize. Laya makes the most sense when you have a stable answer set and representative training data. Fine-tune it for the task, calibrate its probabilities on held-out examples, and send uncertain decisions to a stronger model or human reviewer. Jev gives you stronger zero-shot decisions. Laya gives you speed, privacy, control, and no per-call fee after you teach it your task. The open-source repository is here: http://github.com/NandhaKishorM/laya If you want to self-host Laya, InstaCloud provides a really nice and easy way to do so. Check this out: https://www.instacloud.com/templates/laya If you're new to decision models, I wrote the full breakdown of how Jev. The article is quoted below.

MD Fazal Mustafa
@the_mdfazal
🚨 A LOCAL 421M MODEL JUST ATE CLOUD JEV ON SPEED Laya is an open-source System 1 decision model that runs on your machine. Laya: 86.5 decisions/sec, P50 ~9ms, score 46 Jev: 3.2 decisions/sec, 317ms API round-trip, score 1 Ships with: — 421M params — ~1GB inference memory — Millisecond local calls, no network — Typed decisions in one forward pass — Apache-2.0 and self-hosted While the cloud waits, local decides.

StudioYebisu
@studio_yebisu
やぁ!兄弟たち! Jevに関するGitHubの実用性と発展性がありそうなリポジトリをまとめたよ! やはり、高速判断を要するComputerUseや完全自動トレードなんかに対しての活用が多い印象だね! Jevは公式のウェイトリストも1日ほどで承認されるけど、待たなくてもVercel AI GatewayからModel: typesafe-ai/jevで直接呼び出せるってさ! このGrok翻訳っぽい文章はわざわざタイピングしてるゾ! 1. browser-use/jev-ultrafast Jevがブラウザ自動操作と対象要素を選び、文字入力時のみ小型LLMを使うブラウザエージェント。 https://github.com/browser-use/jev-ultrafast 2. TheoLeeCJ/openjev TypeSafe公式の非公開モデルを再現したものではなく、オープンモデルで同じ入出力パターンを試す独立プロジェクト。ローカルGPUで動かせます。 https://github.com/TheoLeeCJ/openjev 3. vinnylarouge/jevlike 変化するテキスト候補から1つを選ぶJev-likeモデルを、自前データで学習する独立プロジェクト。TypeSafe公式モデルではありません。※使い物にはならないが面白い試み https://github.com/vinnylarouge/jevlike 4. jarrodwatts/jev-trader Monad上のKuru MON-USDCを対象に、Jevで売買判断を行うトレードbot。デフォルトはdry-run/mock動作で、収益性を証明・保証するものではありません。 https://github.com/jarrodwatts/jev-trader 5. Anil-matcha/awesome-jev-by-typesafe TypeSafe Jevの使い方、パターン、プロンプト、スターターコード、事例などを集めたコミュニティのまとめ。 https://github.com/Anil-matcha/awesome-jev-by-typesafe 6. awlevin/typesafe-computer-use 画面をOCRで読み取り、次のクリック操作をTypeSafe/Jevで選ぶMac向けのcomputer-use実装。 https://github.com/awlevin/typesafe-computer-use 7. devagrawal09/jev-review Git diffやコードベースを段階的に検査し、正確性・安全性・信頼性・互換性・テストリスクを構造化して表示するコードレビューworkflow。 https://github.com/devagrawal09/jev-review 8. gargpratyush/jev-router Claude CodeやOpenAI Codexのターンごとに、簡単な作業を速いモデル、難しい作業を強いモデルへ振り分けるルーター。 https://github.com/gargpratyush/jev-router 9. vlad-terin/jev-browser 既存のブラウザ自動操作ツールにJevを組み込み、連続的な観測・操作・検証ループで対象要素を選ぶ非公式runtime/skill。 https://github.com/vlad-terin/jev-browser 10. droidrun/mobile-jev 実機のAndroid端末をJevで操作するモバイルエージェント。ライブ操作デモ、CLI、実行トレースを含みます。 https://github.com/droidrun/mobile-jev
Guidex.com
Jev repositories worth a look, in Japanese
StudioYebisu’s roundup of GitHub repositories, mostly computer use and automated trading.

Baris Terzioglu
@terzioglubrs
Agent memory is usually just an append-only Markdown file that grows forever, and most frameworks load the whole thing into context on every single run. That has been bothering me for a while, but never quite enough to fix it for our agents. Jev feels like it might be a low effort patch to this, by simply scoring each memory against the prompt first, then load only the relevant bits for that specific request.
XContext and memory
Agent memory is usually one growing file

Petru - Tech Driven
@techdrivenpetru
Yes but not with Jev. I used another classifier, locally, just as fast, just as good. Example: setup a "server" in python that loads the classifier model. Add a hook in Claude that fires on "pre-tool-use" and next time you ask Claude a random question like "how do I lint check a project?" and it tries to freelance and read your entire repo only to intoxicate itself and pollute its context, the classifier will slap its hand, say "no sir, you answer from knowledge" and deny the tool call. I tested this yesterday with success, but need to refine it as it misfires. Basically I was able to identify general queries, instances where I would ask something and Claude would rush ahead and run pip install without me asking or just write code instead of answering. A classifier is hypercheap compared to a regular LLM and would catch all of these. The model I used was DeBERT large. It's still stupid fast, I tested it on an Apple with M1 (regular) and you don't feel it running.
XOpen source
A local classifier instead of Jev

たてこ
@tateko_ai
今日のgakuse.aiの勉強会テーマは「Jev」 Jevの基本的な仕組みとか、海外での活用事例など、Jevでできることの広さを知れてめっちゃ勉強になったな! 特に印象に残ったのは、ハーネスにJevを組み込むことで、トークンを効率的に使えるという話 早速Jev使って人狼ゲーム作ってみた!
XAgents and browsers
Jev inside a harness, from a study group

Guidegithub.com
awesome-typesafe
Official resources and community projects for TypeSafe, System One models and Jev.

tanav
@tanavtwt
Created a slop detector extension with Jev It scans all the post on the screen in the real time and classifies it on categories like scam, slop, clean, etc. Shows a minimal badge on the post with the confidence score. Comment bellow if you want to try the extension.
XSocial feeds
A confidence badge on every post