Jev Waiting List: Every Question Answered (2026)

The Jev waiting list ended on 21 September 2026, and this is the full reference for everything people are still asking about it.

I have had the same handful of questions come in over and over since launch week.

How long does approval take, why did nobody email me, is there a way round it, and did I miss the free credit.

So rather than write another opinion piece, I have laid this out as a straight reference — every question with the actual answer and the actual number underneath it.

Nothing here is guesswork about what TypeSafe might do.

Where the company has not published something, I say so rather than inventing a figure.

The timeline in one table

Here is every date that matters, in order.

That is six days from gated early access to open signup.

For a model that fell over under load in its first 48 hours, that is fast.

How the waiting list actually worked

There was no separate form, and that confused a lot of people.

You created an account at console.typesafe.ai, and if TypeSafe had not approved you yet, that unapproved account was your place in the queue.

No position number, no estimated wait, no confirmation screen.

TypeSafe never published how long approval took or how it decided who got in first.

The company said it was moving developers off the waiting list as fast as it could, and that it would let more people in as new GPU capacity arrived.

Reported waits ranged from a few hours to well over a day.

Rate limits were being adjusted dynamically the whole time, because the company was serving far more demand than it planned for.

None of that applies any more, but it explains why so many people are still sitting in an inbox waiting for an email that is never coming.

🔥 Want the exact setup I used to get these results? Inside the AI Profit Boardroom, I've got a full AI automation section with step-by-step video tutorials on wiring models like Jev into real agent workflows. Plus weekly coaching calls + 2,800+ members. → Get access here

What you get now, for free

Every new account gets $5 of credit.

At Jev's input price of $0.042 per million tokens, that is roughly 120 million input tokens.

Output tokens cost nothing at all, because they are so small TypeSafe decided they were too cheap to meter.

To put that in perspective, 120 million tokens of input on a frontier text model would cost you hundreds of dollars.

Here you get it to test with before you enter a card.

TypeSafe has not published credit-expiry terms, so I would not leave it sitting untouched for six months and assume it will still be there.

The full cost picture is in the Jev pricing post.

What Jev is, for anyone who queued without checking

This is the question behind half the confusion.

Jev is a decision model, not a language model.

It takes unstructured state plus a set of questions, and returns typed answers with calibrated probabilities.

It never returns prose, and the output is designed to be consumed by your code rather than read by a person.

TypeSafe calls this a "System One" model — fast, instinctive judgement rather than deliberate reasoning.

There are exactly three question types you can ask.

Choice is capped at 255 options per query.

Picking from thousands of items means scoring candidates in batches first, then running the finalists through one Choice question.

I broke the whole model down in the Jev AI model post and the internals in Jev architecture.

The four ways to reach Jev, compared

Three of these never had a waiting list at any point.

All four serve Jev 1.13, so the answers are identical.

Version pinning is the one that catches people out.

Only the direct API lets you lock to something like jev-1.13.0, and if your thresholds are tuned against a specific version, a silent upgrade on a gateway will quietly change your behaviour.

The gateway breakdown lives in the Jev AI API post.

How fast did adoption actually move

Vercel said Jev became the fastest-adopted model in AI Gateway history.

By hour 24 it was being used by nearly 13% of paid teams, which is more than double the share reached by any previous model launch on the platform.

That is the number that explains the whole waiting list.

TypeSafe priced a decision model at four cents a million tokens and did not anticipate that every team with an agent loop would want to route through it on day one.

The accuracy question, answered honestly

People keep asking whether Jev is "better" than an LLM.

On TypeSafe's own disclosed figures, Jev scores 67.8% aggregate accuracy against 74.1% for a comparable LLM on the same benchmarks.

Its self-published JevBench score is 75.3.

So on raw accuracy, a frontier model still wins.

What Jev wins on is speed and cost, by a margin that changes what you can build.

End-to-end response time is 70ms to 500ms, which TypeSafe puts at 40x to 200x faster than frontier models on System One tasks.

A decision that costs almost nothing and returns in under a tenth of a second can sit inside a loop that runs constantly.

A decision that costs a cent and takes three seconds cannot.

That is the trade, and it is a good trade for routing, triage and gating — and a bad one for anything where a wrong call is expensive.

🔥 Automating real workflows with AI agents? The AI Profit Boardroom has the full automation training — agent stacks, model routing, cost control, and the systems I run inside my own agency. Weekly coaching calls, 2,800+ members. → Join the Boardroom

What people are actually building with it

Model routing is the headline use case.

LangChain shipped a ready-made router built on Jev where you describe each model's strengths in plain English and Jev picks the right one per request.

Under the hood that is a Choice question answered in well under a second, replacing a rules engine nobody enjoyed maintaining.

Support ticket triage fits Score, because severity is an ordered scale rather than a set of categories.

Content moderation fits Noul, because you want a calibrated probability rather than a confident-sounding paragraph.

Lead qualification, spam filtering and agent branch decisions all follow the same pattern.

The test is easy — if you are currently paying a large model to make a small classification, that is a Jev job.

There are more worked examples in the Jev use cases post.

What TypeSafe still has not told anyone

I want to be straight about the gaps, because a reference post that only lists the good bits is useless.

TypeSafe has not clarified whether removing the waiting list means full general availability or an open beta.

There is no standalone rate card beyond the per-token input price.

There are no published rate limit numbers, because they are still being adjusted against available capacity.

There are no stated credit-expiry terms.

The account-approval procedure was never documented while it existed.

And the benchmark figures are self-published, with no independent verification yet.

Plan accordingly — retries with backoff, a fallback path, and your own evaluation set rather than the demo prompts.

Jev waiting list FAQ

Is there a Jev waiting list right now?

No. TypeSafe removed the Jev waiting list on 21 September 2026. Anyone can register at console.typesafe.ai and create an API key immediately, with $5 of free credit attached to the account.

How long did Jev waiting list approval take?

TypeSafe never published an approval time or the criteria it used. Reported waits ran from a few hours to more than a day, and roughly 140,000 signups were cleared within about 36 hours before the queue was removed.

Why did I never get an approval email for Jev?

Because the queue was switched off before your turn came round. There is no email to wait for now — log back into console.typesafe.ai with the email you registered and create a key directly.

Was there ever a way to use Jev without joining the waiting list?

Yes. OpenRouter (typesafe/jev-1.13), Vercel AI Gateway (typesafe-ai/jev) and Cloudflare Workers AI (typesafe/jev) all served the same model with no approval step throughout early access.

How much does Jev cost after the free credit?

Input is $0.042 per million tokens, which is $42 per billion. Output tokens are free. The $5 starting credit covers roughly 120 million input tokens.

Is Jev accurate enough to replace an LLM for classification?

On TypeSafe's own figures Jev scores 67.8% aggregate against 74.1% for a comparable LLM. It is a good fit where speed and cost matter more than the last few points of accuracy, such as routing and triage, and a poor fit where a wrong decision is expensive to undo.

About Julian

I'm Julian Goldie — AI entrepreneur, SEO expert, and founder of the AI Profit Boardroom (2,800+ members). I help business owners scale with AI agents, automation, and SEO.

• 400K+ YouTube subscribers
• 7-figure AI agency (Goldie Agency)
• Daily training inside the Boardroom
• Author of multiple AI automation playbooks

→ Get my best AI training inside the AI Profit Boardroom

Also On Our Network

• 🌐 Read on bestaiagentcommunity.com
• 🌐 Read on aiprofitboardroom.com
• 🌐 Read on aisuccesslabjuliangoldie.com
• 🌐 Read on aimoneylabjuliangoldie.com

Related reading

• How To Use Jev: State, Questions, Decisions — the practical setup walkthrough.
• Jev AI API — every access route compared in detail.
• Jev Pricing — what $0.042 per million works out to at real volume.
• Is Jev Open Source? — what TypeSafe released publicly.

If you still had a tab open waiting on the Jev waiting list, close it — the queue is gone and your key is three minutes away.

📺 Video notes + links to the tools 👉

🎥 Learn how I make these videos 👉

🆓 Get a FREE AI Course + Community + 1,000 AI Agents 👉