A browser agent that finds flights in seven seconds. A Claude session cut from a million tokens to 86,000 in one second. A WHERE clause that reads plain English. Vercel's safety reviewer. We are collecting what people built with Jev in its first days, and what they argued about, each in their own words. Nobody in these bylines is on eChai, and every entry links back to where it was posted.
Diogo Almeida is the CEO of TypeSafe AI, the San Francisco lab that came out of stealth with this post on 15 September, and before that a researcher at OpenAI on the instruction-following work that led to ChatGPT. Jev is the lab's first model and it is not a chatbot: you hand it the state of a program plus typed questions, and it answers every question at once as a decision with a calibrated probability, in 70 to 500 milliseconds, with no text generated at all. TypeSafe prices it at $0.042 per million input tokens with output free, and its own evals put it up to 193.6x faster and 444.6x cheaper than GPT-6 Astra and Fable 5.1 on four workflows the lab wrote itself, a caveat the launch post lists. The company raised a $40 million seed led by DCVC. The film is the launch video, and the second post in the thread is Jev playing Doom. Nearly everything else on this page was made in the four days after.
Diogo Almeida
2w
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
Andre Landgraf's showcase of the yes/no gate, several worked examples at the link.
Andre Landgraf
2w
Ok, jev is really cool. So many great use cases where a fast and accurate yes/no gate can deliver meaningful performance gains.
Played around and built some showcase examples here →
Vipul Sharma's Minecraft agents ask Jev whether to fight or run, from what the agent can see, the mob count, its weapon and its health; one broke off mid-fight as the odds turned.
Vipul Sharma
2w
I hooked up Jev by @typesafeai to a few agents on my Minecraft world. Look at how Jev decides when they should fight and when they should run away.
Jev rates the odds from what the agent can see, mob count, weapon, how hurt it is, and runs away when they are bad. One broke off mid-fight as its odds turned.
Mahmoud's simulation of fifteen autonomous drones through an asteroid field, Jev deciding for each in under 300 ms, all fifteen arriving.
Mahmoud
2w
Jev can control a swarm of 15 autonomous drones in real time!
I built a simulation to test it navigating an asteroid field:
• Decision latency: < 300ms
• Survival rate: 100% (all 15 reached the destination)
Built with @typesafeai by @CompleteSkeptic
Matthew Berman's second video, eleven minutes, three days in. It plays where it was posted.
Matthew Berman
2w
We need to talk about Jev
Higgsfield, whose API fronts fifty generative models, has Jev pick the most cost-effective model for each request.
Higgsfield
2w
We found the perfect use case for Jev: auto-routing GenAI models.
The idea: Jev picks the most cost-effective models for your task, balancing speed and quality.
All generations made with Higgsfield API.
Chris Tate, of Vercel, on json-render with Jev deciding the tree: generative UI rendered in milliseconds from your own components and design system. Guillermo Rauch's reply was that Jev generative UI had been achieved externally.
Chris Tate
2w
New experiment: json-render + jev
The future Generative UI is instant
Your components, your actions, your design system
Rendered in milliseconds
Borja, of Distribb, on internal linking as a Jev job: 586 pages read and the link map rebuilt in 45 seconds for $0.21, with 139 pages it declined to link because nothing fit, against Claude Opus 5 getting through 21 pages for $1.43 on the same clock. His framing is the useful part: it is not writing, it is 8,790 yes/no calls.
Borja
2w
Jev is WILD for SEO audit 🤯
in 45.1 seconds it read all 586 pages on my site and rebuilt the internal link map. 584 links placed, 139 pages it refused to link because nothing honestly fit. total cost $0.21.
Claude Opus 5, same 586 pages, same clock, got through 21 of them and spent $1.43.
per page that is ~190x cheaper. the full Opus pass would have run $43.
internal linking is the perfect Jev job. it is not writing, it is 8,790 yes/no calls: does this page have a real reason to link to that one, and is there anchor text already sitting in the copy. that is a classification problem, and we have been paying frontier prices to do it one page at a time.
what you are watching: left column is Jev, right is Opus, same queue, same rubric. the run stops the moment Jev finishes so Opus stops burning tokens.
coming soon to @distribb_io + connector + gpt plugin
Greg Isenberg, of Late Checkout, with the businessperson's explanation: Jev is the sorting part of a job, and a great many jobs are exactly that. The list of what it unlocks runs from instant quotes to inbound triage, and the episode of Startup Ideas with Ryan Vogel is linked. The film is 28 minutes and plays where it was posted.
Greg Isenberg
2w
Jev is HERE and this is the CLEAREST explanation of what it is and what NEW businesses it unlocks.
(and at the end I'll tell you how to get Jev even if you're on the waitlist)
WHAT IT IS
You know how you open your inbox and have to decide what's junk, what needs a reply, and what can wait? Jev does that part. It looks at each thing and says "this is junk, I'm 94% sure."
It doesn't write anything back to you. It just sorts.
1,700 emails for 18 cents, instantly.
That sounds kinda trivial but the important part
WHAT IT UNLOCKS
My explanation of Jev sounds small until you realize HOW MANY jobs are exactly this. Someone reading a stack of applications. Someone deciding which support ticket goes to which team. Someone looking at inbound and deciding who's worth calling back.
A few ideas on what it unlocks:
1/ Instant quotes that are actually instant. Every quote form on the internet says "we'll email you by end of day." Build the version that answers in under a second, for roofers, movers, insurance, legal intake.
2/ Lead scoring as a product. Every agency and service business has a contact form full of junk. Score every submission and send the real ones straight to the owner's phone.
3/ Support triage for companies with no support team. The ticket gets classified and routed before anyone opens it.
4/ Clipping tools. Pass in a transcript, get the best moments scored in three seconds. Every clipping product just got a cheaper engine.
5/ Application piles. Grants, permits, insurance claims, job apps, loan docs. Someone reads that stack one item at a time today.
6/ Marketplace matching. Someone types what they need and gets matched to the right local business instantly instead of waiting for callbacks.
7/ Browser agents that actually move FAST. That makes bulk browser work practical: pulling quotes from five carriers, filing the same form for 200 clients, checking supplier inventory in real time etc.
TLDR; find an expensive queue and put Jev at the front of it.
HOW TO GET IT
I didn't realize you can skip the waitlist because Jev is live on the Vercel AI Gateway right now, so you can start calling it today. In this episode, we share how.
Episode now live on @startupideaspod (thanks to @ryanvogel for coming on and spilling the sauce today)
Watch: https://www.youtube.com/watch?v=4mTLpuQpB80
Jev is a big deal because this is a whole new way to do AI
Really cool
Happy Jev day.
Taras builds Flowsery, a session-replay tool. Three million replay events: in 40 seconds Jev watched 3,247 sessions, flagged 132 rage clicks, 116 dead clicks and 95 JavaScript errors, and opened 213 draft pull requests, for $2.17. The numbers are the author's.
Taras
2w
JEV is INSANE.
We gave it 3 million replay events
In 40 seconds, it watched 3,247 sessions, caught 132 rage clicks, 116 dead clicks and 95 JavaScript errors, and opened 213 draft PRs to fix them.
All for just $2.17.
JEV can also flag abandoned forms, rank every issue by severity, count how many sessions each bug hit and write the repro steps for your team.
Coming soon to Flowsery + MCP.
Comment "JEV" for early access.
Matt Mastracci's djev: DiffusionGemma made to answer like Jev, running near-real-time vision detection on a phone through the model's own vision tower.
Matt Mastracci
2w
DiffusionGemma-as-Jev (aka djev) running near-real-time vision detection from a mobile phone using its native vision tower.
Please don't fall down the stairs!
Hassan El Mghari's chess match: GLM 5.3 won in 29 moves, Jev played each move in about 0.3 seconds for under a hundredth of a cent against GLM's 5.8 seconds and 0.8 cents, and the whole game cost 24 cents. His takeaway is the hybrid pattern he had already used for fraud detection.
Hassan El Mghari
2w
Jev vs GLM 5.3 at chess!
Results:
◾ GLM 5.3 won by checkmate in 29 moves
◾ Jev: ~0.3s and <$0.0001 per move
◾ GLM 5.3: ~5.8s and ~$0.008 per move
◾ The whole game cost 24 cents
My main takeaway is that it's often useful to use each one to their strengths:
◾ Fast, well-defined classification → specialized models like Jev
◾ Classifications that need reasoning or lookahead → LLMs like GLM 5.3
◾ Real classification pipelines → hybrid. Jev handles the easy calls, an open model handles the hard ones.
The future is multi-model!
Mahesh Sathiamoorthy runs Bespoke Labs. Nimble is an open Jev: open data (fully synthetic, curated by slightly altering facts to make negatives), an open recipe (a LoRA fine-tune of Qwen 3.5 9B, Jev used only to evaluate, no RL yet), an open model on Hugging Face, and parallel constrained decoding to serve it. He states the caveat himself: there is no standard benchmark, so Nimble may be much worse than Jev elsewhere.
Mahesh Sathiamoorthy
2w
Introducing Bespoke Nimble: an open data, open model, open recipe for an open Jev.
Code and info: https://github.com/bespokelabsai/nimble
Model: https://huggingface.co/bespokelabs/Bespoke-Nimble-9B
Data:
* A new data curation recipe called contrastive data curation.
* Slightly change facts to generate negative data. This pushes the model to discriminate better and become a better decision maker. The calibration is implicit.
* Didn't do ablations but I think this is a critical piece!
* This also means training data doesn't need probabilities.
* Data covered 10 categories, and is fully synthetic.
* This data is split into train and eval.
Training
* LoRA finetune of Qwen3.5-9B.
* Distillation-free: we use Jev to only evaluate.
* No RL yet!
Serving
* Parallel constrained decoding as suggested by @NielsRogge and @harshagundal.
Results:
* The post-trained Qwen (Nimble) became substantially better on our curated eval: 66% for Qwen to 90% for Nimble. Jev is at 93%.
* 100ms on H100 and free to use on your macbook! Feel the AGI for free.
* 2 days of building in public. :)
Big caveat is that there is no standard benchmark to measure performance, and it's possible Nimble is much worse on other benchmarks compared to Jev. But it should be better than Qwen!
We thank @typesafeai for making Jev and the inspiring discussions in the community. Hope this release lifts all the boats and encourages more research and activity in this space.
Jack Cheng's second demo, on the Bluesky public firehose: every English post, first whether it mentions an animal and then whether that is a real animal, a sports team or a figure of speech, close to real time.
Jack Cheng
2w
Another demo to show Jev’s speed. Here it takes all the english language posts from the Bluesky/AT protocol public firehose and 1) figures out which have an animal in them and 2) classifies whether it’s an actual animal, sports team, a figure of speech, etc, pretty much real-time
Ian Nuttall on the quickest way in: TypeSafe publishes a skill, one command installs it, and a prompt in your own project asks where Jev could replace a slow model call.
Ian Nuttall
2w
Unsure how to get started with Jev?
Install the skill:
npx skills add typesafe-ai/skills --skill typesafe-ai
Then prompt in your project:
use /typesafe-ai to see how Jev can be used to replace slow, expensive LLM usage and find possible new features it would enable for users.
