Jev Waiting List Is Gone (Free $5 Credit)

The Jev waiting list no longer exists — TypeSafe AI removed it on 21 September 2026, so you can sign up right now at console.typesafe.ai and start making calls today.

That is the whole answer, and I wanted it in the first line because I know exactly why you are here.

You joined the queue, you have been refreshing your inbox for days, and nobody told you the gate came down.

It did come down, six days after launch, and roughly 140,000 signups were cleared inside 36 hours.

Every new account gets $5 of free credit, which works out at about 120 million tokens on Jev's pricing.

This post covers what changed, what Jev actually is, how to get your API key in about three minutes, the three no-waitlist routes that already existed, and the limits worth knowing before you build anything real on it.

What happened to the Jev waiting list

Jev launched on 15 September 2026 in limited early access, and the only way in was a queue.

You created an account at console.typesafe.ai, and if you were not approved yet, that account sat in the waiting list by default.

There was no separate form and no confirmation page telling you where you stood.

TypeSafe never published an approval time or explained how it decided who got in first, which is why the whole thing felt so opaque.

Demand was heavy enough that the API briefly stopped serving requests during launch week.

Rate limits were being adjusted on the fly while the company waited on new GPU capacity.

Then on 21 September the queue was switched off entirely.

Anyone can now register and get a key immediately, with $5 of credit attached to the account.

🔥 Want the exact setup I used to get these results? Inside the AI Profit Boardroom, I've got a full AI models and APIs section with step-by-step video tutorials covering Jev, routing and agent stacks. Plus weekly coaching calls + 2,800+ members building real automations. → Get access here

What Jev actually is, in one sentence

Jev is a decision model from TypeSafe AI that takes messy state plus your questions and returns typed answers with calibrated probabilities, instead of returning text.

That last part is the bit most people miss when they join the queue.

Jev does not write prose.

It does not chat, it does not draft emails, and it will never produce a paragraph for you.

TypeSafe calls it a "System One" model, borrowing the fast-instinctive-thinking idea from psychology.

You hand it the state of your system and a list of questions, and it hands back structured decisions your code can consume directly.

A human is not supposed to read the output at all.

TypeSafe is a San Francisco company founded in 2024, and it announced a $40 million seed round led by DCVC on the same day Jev launched.

If you want the deeper architecture breakdown, I covered it in the Jev AI model explainer and the Jev architecture post.

The three primitives you get

Everything Jev does comes down to three question types.

Primitive	What you ask	What comes back
Choice	Pick one option from a list you supply	A probability for every option, plus an overall confidence number
Score	Rate the state against ordered levels you define	A continuous score plus a distribution across your levels
Noul	A straight yes or no	A single probability that the answer is true

That is the entire surface area.

Choice caps out at 255 options per query, so if you are picking from thousands of items you score candidates in batches first and then run the finalists through one Choice question.

The reason this matters for the waiting list conversation is simple.

Plenty of people queued up expecting a cheaper GPT, and what they got access to is a completely different shape of tool.

Knowing that before you burn your free credit saves you an afternoon.

How to get Jev access now (the three-minute version)

Here is the exact path, and none of it involves a queue any more.

First, go to console.typesafe.ai and create an account with your email.

Second, confirm the email and open the console — you should land straight in rather than on a holding page.

Third, create an API key from the keys screen and copy it somewhere safe, because you will not see it again.

Fourth, set it as an environment variable called TYPESAFE_API_KEY rather than pasting it into your code.

Fifth, install the official SDK for your language, or the langchain-typesafe package if you are already inside a LangChain stack.

Sixth, send a tiny request with one Noul question so you can confirm auth works before you write anything complicated.

Your $5 credit covers roughly 120 million input tokens, which is an enormous amount of testing for a model that charges nothing for output.

I walked the whole flow end to end in the how to use Jev guide, and the free-credit angle specifically in how to use Jev AI for free.

The routes that never needed a waiting list

This is the part that annoyed a lot of people when they found out.

While the direct queue was running, the same model was already available through three gateways with no approval step at all.

Route	Model ID	What you need	Waitlist
TypeSafe direct	`jev-1.13`	Account at console.typesafe.ai	None now
OpenRouter	`typesafe/jev-1.13`	Prepaid OpenRouter credits	Never had one
Vercel AI Gateway	`typesafe-ai/jev`	Card on the team	Never had one
Cloudflare Workers AI	`typesafe/jev`	Cloudflare account	Never had one

All four serve the same Jev 1.13 model, so the output is identical.

Going direct still buys you a few things the gateways do not offer.

You get the official Python and JavaScript SDKs rather than a generic client.

You get version pinning, so you can lock to something like jev-1.13.0 and stop a silent upgrade breaking your thresholds.

You get the status page and a larger documented request budget.

If you are shipping something to production, go direct.

If you just want to poke at it inside an app you already host on Vercel, the gateway is two lines of config and you are done.

I broke the gateway options down properly in the Jev AI API post.

What it costs once the free credit runs out

The pricing is the reason people queued in the first place.

Input runs at $0.042 per million tokens, which is $42 per billion.

Output is free, because the outputs are so small TypeSafe decided they were too cheap to meter.

End-to-end response time sits between 70ms and 500ms.

TypeSafe claims that is 40x to 200x faster than frontier models on the same System One tasks.

That combination is the actual product — not raw intelligence, but decisions cheap and fast enough to put inside a loop that runs thousands of times a minute.

Full numbers and worked examples are in the Jev pricing breakdown.

🔥 Building agents that need fast routing? The AI Profit Boardroom has the full agent-stack training — model routing, cost control, and the exact workflows I run in my own agency. Weekly coaching calls and 2,800+ members inside. → Join the Boardroom

The honest limitations

I am not going to pretend the numbers are all flattering.

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

Its self-published JevBench score is 75.3, and self-published benchmarks are self-published benchmarks.

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

There is no standalone rate card beyond the token price, and no published credit-expiry terms.

Rate limits still move around based on capacity.

So test it against your own production data rather than the demo prompts, and keep a fallback path in your code for the days the API wobbles.

That is not a reason to skip it — it is a reason to wire it in carefully.

Where Jev is genuinely worth using

The clearest win I have seen is model routing.

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

Under the hood that is just a Choice question answered in well under a second, and it replaces a hand-written rules engine nobody wanted to maintain.

Content moderation is the second obvious one, because a Noul question with a calibrated probability is exactly what you want there.

Support ticket triage fits neatly into Score, since severity is an ordered scale rather than a category.

Lead qualification, spam filtering and branch decisions inside an agent loop all fall into the same bucket.

The pattern is always the same — something your agent currently asks a big expensive model to decide, that is really just a classification.

More worked examples live in the Jev use cases post.

How fast Jev got adopted

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 any previous model launch reached.

That number tells you why the waiting list existed at all.

It also tells you the pressure that forced TypeSafe to drop it within a week.

When a model is cheap enough to sit inside every request path, adoption does not look like a normal launch curve.

Jev waiting list FAQ

Is there still a Jev waiting list in 2026?

No. The Jev waiting list was removed on 21 September 2026, six days after launch. You can create an account at console.typesafe.ai and generate an API key immediately, with $5 of free credit attached.

How long did the Jev waiting list take?

TypeSafe never published an official approval time. Early reports ranged from a few hours to more than a day. Roughly 140,000 queued signups were cleared within about 36 hours before the queue was removed entirely.

Do I still need to join the Jev waiting list to use the API?

No, and you never strictly did. OpenRouter, Vercel AI Gateway and Cloudflare Workers AI all served the same Jev 1.13 model without any approval step throughout the early access period.

How much free credit do I get with a new Jev account?

Every new account gets $5 of credit, which is roughly 120 million input tokens at Jev's rate of $0.042 per million. Output tokens are not charged at all.

Is Jev a replacement for GPT or Claude?

No. Jev does not generate text. It returns typed decisions — choices, scores and yes/no answers — each with a calibrated probability. You keep a generator model for prose and use Jev for the routing and verification around it.

Can I pin a Jev model version?

Yes, but only through TypeSafe's direct API. Version pinning to something like jev-1.13.0 is one of the things you lose if you access Jev through a gateway instead.

My verdict on the whole waiting list saga

The queue was a capacity problem, not a marketing stunt.

TypeSafe underestimated how many teams would want a decision model priced at four cents a million tokens, and the API fell over during launch week because of it.

Six days later the gate came down, which is about as fast as anyone could reasonably expect.

If you are still sitting in an inbox waiting for an approval email, stop waiting and go create a key.

The $5 credit is enough to prototype an entire routing layer without spending anything.

And the honest advice is to test it on your own data before you rip out whatever you are currently using — 67.8% aggregate accuracy is good enough for routing and triage, and not good enough for anything where a wrong call is expensive.

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.