X postGames and real time
Jev plays Doom
A game loop that asks a model what to do roughly ten times a second.
by TypeSafe AI
- Rate
- ~10 queries/s
- Cost
- ~$7/hour
A directory of Jev projects
What people are building with Jev
Jev is the first System One model from TypeSafe AI: it gives up writing text and returns typed decisions with calibrated probabilities, in well under a second. This is the running list of what that has been used for, with the cost and latency each author reported.
- Response time
- 70–500 ms
- Input tokens
- $0.042 / MTok
- Choice ceiling
- 255
State
“Hi, I’ve been trying to connect my Stripe account for 3 days and it keeps failing. I’m losing sales. Please help ASAP.”
departmentchoice
- billing0.840
- technical0.159
- sales0.001
Confidence 0.60 · answered in milliseconds
Example from the TypeSafe quickstart.
Browse by what it decides
Seven shapes of problem, from a 300 ms trading loop to a corpus of a thousand papers.
- Agents and browsers4Loops that pick the next action: click a link, call a tool, hand off to a human.
- Games and real time3Decisions inside a frame budget, where a reasoning model arrives too late to matter.
- Triage and routing1Inboxes, tickets and queues sorted into buckets a confidence threshold can act on.
- Trading and markets1Judgment on a price feed, at the speed the market moves.
- Content and growth4Scoring drafts, posts and ads before they ship, one question per attribute.
- Research and data1Map-reduce over a corpus: classify every document, then read the aggregate.
- Robotics and devices1Control loops on hardware, where latency is a physical constraint.
Every build
15 projects · updated Sep 18, 2026
- ArticleGames and real time
Wikiracing
Pick one link out of thousands, then do it again until you arrive.
by TypeSafe AI
- Choice ceiling
- 255 options
- GitHub repoAgents and browsers
jev-ultrafast
Browser Use’s agent, with the next-action decision moved to Jev.
by Browser Use
- Stars
- ~2.9k
- X postAgents and browsers
Flight search with Browser Use
A booking flow driven end to end in about seven seconds.
by Gregor Zunic
- Run time
- ~7 s
- Cost
- ~$0.004
- X postAgents and browsers
Stagehand on a remote browser
Browser tasks at about a tenth of a cent each.
by Kyle Jeong
- Cost per task
- ~$0.001
- GitHub repoAgents and browsers
typesafe-computer-use
Computer use on macOS, one typed decision per step.
by Andrew Levin
- GitHub repoTrading and markets
jev-trader
Buy or sell, decided inside a 300 ms block.
by Jarrod Watts
- Decision window
- 300 ms blocks
- GitHub repoRobotics and devices
jev-drone
A tactical judgment loop flying on hardware.
by Roman Slack
- X postTriage and routing
Triage across 1,500 emails
A full inbox sorted in one pass.
by Ryan Vogel
- Volume
- ~1,500 emails
- ArticleContent and growth
Every’s editorial vibe check
Eleven experiments, 1,709 judgments, under a cent in total.
by Mike Taylor
- Judgments
- 1,709
- Total cost
- <$0.01
- Median per passage
- 0.35 s
- Live siteResearch and data
1kpapers
A thousand papers sorted by topic, then published as a site.
by 1kpapers
- Documents
- 1,018 papers
- ArticleGames and real time
Jev plays chess
Legal moves as a Choice, scored against reasoning models.
by Maxim Saplin
- X postContent and growth
3,282 posts, eight questions each
A back catalogue scored to find what actually travels.
by Ian Nuttall
- Posts
- 3,282
- Cost
- $0.1282
- Run time
- 8 m 34 s
- X postContent and growth
Post scoring with SuperX
61 questions about a draft, answered in about a second.
by Rob Hallam
- Questions per draft
- 61
- Latency
- ~1 s
- Cost
- $0.0004
- X postContent and growth
724 competitor ads, broken down
Every hook, format, offer and call to action, in forty seconds.
by Matthew Berman
- Ads
- 724 from 37 brands
- Run time
- ~40 s
- Cost
- ~$0.09
In their own words
The posts these entries come from, quoted as published.
Jarrod Watts
@jarrodwatts
I built a trading bot with Jev! Jev decides if it should “buy” or “sell”, given the price feed of an asset pair, and executes real trades. It uses Monad to place the orders on Kuru’s on-chain order book in every 300ms block.
Sep 16, 2026Read the post Ian Nuttall
@iannuttall
I gave Jev 3,282 of my X posts across 100M views and asked it to find what actually works for growth. 4,252,330 tokens $0.1282 for the full 8m 34s run! Each post got 8 questions about the topic, hook, tone, whether it teaches something, etc.
Sep 17, 2026Read the post Rob Hallam
@robj3d3
Jev + SuperX = virality solved ✅ Every post gets 61 questions in ~1s for $0.0004 🤯 > fitted on 9,481 real posts from 207 creators > picks the viral post 2 in 3 times > never rewards reply bait So: write, score, rewrite, stop when it peaks. Free, no signup. try it below ↓
Sep 17, 2026Read the post Matthew Berman
@TheMattBerman
jev is INSANE. in 40 seconds it broke down 724 live ads from 37 brands. every hook. every format. offer. cta. awareness stage. landing page mismatch. used 9 cents of tokens. (will be avail in @stealads + mcp)
Sep 17, 2026Read the post
Guides and references
Worth reading before you wire Jev into anything.
- Introducing System One Models and Jev
The launch post: what the model class is, the evals, and the caveats TypeSafe put on them.
typesafe.ai
- Quickstart: state, questions, typed answers
The three question types — Choice, Score and Noul — in one support-ticket example.
docs.typesafe.ai
- Workflow evals
Accuracy, cost and time per case across four automation workflows, model by model.
evals.typesafe.ai
- Jev on Cloudflare
Worked examples for support routing, refund decisions and risk escalation.
developers.cloudflare.com
- Building a harness with Jev
LangChain on model routing and gating dangerous tool calls behind a typed decision.
langchain.com
- Jev, from a developer’s angle
Flavio Copes on triage, RAG filtering, citation checks and confidence gates.
flaviocopes.com
- awesome-jev
Community list of projects, wrappers and examples.
github.com
- The launch thread on Hacker News
Where the sceptical reading of the benchmarks lives.
news.ycombinator.com
Built something with Jev?
Send me a DM on X with a link. If it shows something the list does not cover yet, it goes up with your name on it.