Case wall
Jev AI examples, page 15
407 real Jev AI builds and public demos, updated daily. Original text, metrics, and a link back to each source.
Last updated
Public Jev AI examples and demos, not official TypeSafe docs. Clients and SDKs are in the Jev GitHub projects. Paths for jev-latest, Vercel, and OpenRouter are in the September 2026 channel map.
Agent-skill repositories are listed on Jev skills on GitHub.
OpenRouter model ids recorded for Jev AI are on Jev AI on OpenRouter. Rubric and eval posts are on Jev as a judge. How the company and the model fit together is on TypeSafe AI. The pricing calculator uses the published input rate.
No posts match these filters.
@a1exstone explains Jev as a System One model: send state and questions, get typed answers with probabilities, with questions evaluated in parallel. They recount a two-turn agent loop where six of eight decisions (pick a model, screen a tool call, check if done) cost 0.06s combined, and cite TypeSafe figures of 20–200x faster and 40–400x cheaper on classification.
TYPESAFE RELEASED A MODEL THAT DOESN'T GENERATE TEXT, AND IT BREAKS THE COST FLOOR OF THE AGENT LOOP jev is what they call a system one model. you send it a sta…
Quoted @sydneyrunkle
@ziwenxu_ explains calling Jev on OpenRouter while TypeSafe still has a waitlist: sign up, add credits, make a key, and POST /api/alpha/decisions with model typesafe/jev-latest, a state, and questions. They cite $0.042 per million input tokens and free output, a billing example at 100% in about a second, and Tetris moves in about half a second.
Jev is already on OpenRouter!! It's only been out a day and TypeSafe still has a waitlist. OpenRouter has it though, and that's how we've been running it. The p…
@vladdubchak_x built a Maxfusion plus TypeSafe tool that scores every shot of a Meta video ad in 1.5 seconds. They pushed 450-plus ads through in under 3 minutes, and say one Duolingo ad received 50 judgments in 1.52 seconds for $0.00028 and a score of 88.
Jev is insane for ad research. I built a tool that scores every shot of any video ad on meta in 1.5 seconds. It runs on Maxfusion + @typesafeai I pushed 450+ ad…
@chiziaruhoma gave an evolution simulation to Jev. Two species with opposite DNA ran 14 generations through an ice age, 432 creatures, and Jev decided who survived, who mated, and what killed each one.
I gave an evolution simulation to Jev, a small model from @typesafeai that answers typed questions with odds instead of writing text. Two species with opposite…
@marcelpociot built Wordshift, a semantic typing-racer with TypeSafe Jev. Comparisons such as “slower than a jet, bigger than an elephant, faster than a cheetah” move the car farther when the difference is bigger.
Jev opens up so many possibilities for gaming 🤯 🏁 I built Wordshift - the first semantic typing racer game built with Jev from @typesafeai Slower than a jet, bi…
@ephraimduncan had TypeSafe Jev play Pac-Man in real time by choosing a direction at each junction from the maze state as JSON. The demo is at jev-pacman.ephraimduncan.com.
I got access to @typesafeai's Jev model and got it to play Pac-man Jev decides which direction Pac-Man should turn at each junction, given the maze state as JSO…
@princecaarlo runs an agent-browser and Jev loop for browser use. The stated goal is a Wikipedia race from Coffee to Artificial Intelligence.
okay so browser use is a legitimate use case agent-browser <> Jev loop goal: wiki race from 'Coffee' to 'Artificial Intelligence' https://t.co/565VqWYgoJ
Quoted @princecaarlo
@yelkhayami built an extension that judges tweets as they are typed. They say it reacts to each keystroke and that an hour of testing cost half a cent.
i got early access to jev by @typesafeai and decided to build a little extension that judges my tweets while i type them it runs so fast it can react to each ke…
@rrriviannn asked whether TypeSafe Jev could generate images and posts a sunflower “according to Jev.” The text does not describe the method.
I was wondering if @typesafeai Jev could generate images so here's what a sunflower looks like according to Jev : https://t.co/vT4UiV4SqJ
Quoted @CompleteSkeptic
@tosa_now is trying product ideas with Jev. One is drag-and-drop that tells you the sort destination automatically: the first call is a bit slow, and later calls feel snappy.
Jev (@typesafeai) 使って色々プロダクトで使えそうなの考えてみてる。 dndをしたら自動で仕分け先を教えてくれる体験 最初一回はちょっと遅いが、そのあとはサクサク https://t.co/2Z7K8FA9XI
@framer uses TypeSafe as an example of a team that still comes to Framer when a site has to look like a serious company on day one. The post is a Framer advertisement, not a Jev build.
Building got faster. Taste didn’t become optional. Framer is where frontier teams like @typesafeai still go when the site has to look like a serious company on…
Quoted @CompleteSkeptic
@JustLingonberry shows TypeSafe playing Minecraft for 2 minutes at 1 cent (150k tokens). They say it dodges zombies at nightfall with no prompting.
@typesafeai playing minecraft, costing only 1 cent (150k tokens) for 2 minutes of play automatically knowing to dodge zombies at nightfall with no prompting htt…
Quoted @thread
@gabrycina gave Jev a robot arm whose goal was to put a red cube in a green box. The cube was out of reach, so the model decided to grab a hook, drag the cube closer, put the hook down, and finish. They say tool use was not programmed.
I gave Jev a robot arm. Goal: red cube in the green box. Catch: the cube was out of reach. It grabbed a hook, dragged the cube closer, put the hook down, and fi…
@ojusave built a hot-dog or not hot-dog classifier, referencing Silicon Valley, powered by TypeSafe and Render workflows, at seefood-6wov.onrender.com.
I gave @typesafeai the most important binary in @SiliconHBO history. hot dog or not hot dog Try here: https://t.co/m6LmF6svsB Powered by @render workflows https…
@Entelic_Aria describes Jev sending each task to the model that should take it, and says Jev decisions cost $0.04 per 1M input tokens on Aria with no markup.
Every model wants the job. The fast one says, “I’ve got this.” The reasoning model asks for more context. The safety model says, “Absolutely not.” Jev steps in…
@thiagobrez posts a live TypeSafe Jev demo at AgentConf. The text does not describe what the demo does.
Live @typesafeai Jev demo at @AgentConf https://t.co/y3UpKP2e4z
@lomeshdutta open-sourced a Skill Router built with Claude and TypeSafe Jev. They have about 90 Claude Code skills and use maybe five, so the router is meant to surface a skill Claude would not call on its own.
Just hacked a fun little project - Skill Router (Open source) using Claude and Jev @typesafeai Why? I have ~ 90 Claude Code skills installed and use maybe five.…
Quoted @CompleteSkeptic
@iamzhe says minirouter/fusion is live: it classifies each request with TypeSafe Jev, then routes to a model matched to type and complexity. Codex lists their presets and routers in the model picker. They also fixed long prompts not caching automatically in Claude Code.
Small update for https://t.co/xXR4JLt377 today: 1. minirouter/fusion is live. It classifies each request with Jev by @typesafeai, then routes to a handpicked mo…
@imarikchakma built a syntax highlighter with TypeSafe Jev that they say can highlight any language, including one they made up.
Built the ultimate syntax highlighter using @typesafeai Jev (i mean why not). It can highlight any language, even one that I made up.
@kzkhykw used TypeSafe Jev to judge local Whisper transcripts and automate a Mac. Chrome pages use a browser harness; Slack and Notion are seen and operated through Mac accessibility. They say the implementation was done with Grok 4.6 only.
. @typesafeai のJevを使って、Local Whisperの文字起こしを都度判定して、Macbookのローカルコンピュータの操作を自動化して、あれこれするのやってみました！ Chrome のページは Browser Harness、Slack やNotionは Mac の Accessibility で認…
Quoted @kzkhykw
@iamAdityaAnjana used Jev to write C++ by giving it only the grammatically valid next tokens at each step and letting it pick one. They got a working fragment, and say it is slower and worse than a code model and is a novelty, not a recommended way to write code.
Ran a dumb experiment last night: used Jev by @typesafeai (a decision/classifier model, not a text-gen model) to write C++. At every step I gave it the finite s…
@the_cyw says they are building egocentric training data for physical AI and that Jev QA’d 58,643 action labels in under 3 minutes for 90 cents.
Jev is insane. We are building egocentric training data for physical AI. Jev QA'd 58,643 action labels in under 3 minutes. 90 cents. A friction cost of the Clau…
@Juris_Savos shows Jev playing Balatro. Each hand it sees the cards, jokers, and blind, then picks play or discard in about 200–500ms, and shops after the ante. They say a full run costs less than a cent.
Jev from @typesafeai plays @BalatroGame. Every hand it sees the cards, jokers, and blind, then picks play or discard in ~200–500ms. After the ante, it shops too…
@adrianmg switches a game between manual and Jev-driven play at 2–3 calls per second at 60 FPS, with about 150ms average latency. They say the interesting part was turning the game into variables, actions, and criteria, such as whether to risk a powerup or protect the ball.
Had some fun taking Jev from @typesafeai for a spin, switching seamlessly between manual and AI-driven gameplay. It’s clocking 2–3 calls/sec at 60 FPS, with ~15…