Jev Waitlist: There Is No Queue Now, And Here Is Every Way In

There is no Jev waitlist to join right now, because TypeSafe AI reopened sign-ups on 27 September 2026 and you can create an account at console.typesafe.ai without waiting for approval.

You can also skip the account completely, because four other routes serve the same model with no queue.

I re-checked every route on 11 October 2026, so this is the current picture and not launch-week news.

Below you will find what Jev is, what happened to the queue, how to get in, and what to do once you are in.

The Jev waitlist answers in one table

These are the four questions I get asked, with the short answers.

What Jev is, and who makes it

Jev is a decision model made by TypeSafe AI, and it launched in early access on 15 September 2026.

It does not write text.

You send it a "state", which is the email, ticket or log you want judged, plus one or more typed questions.

It returns typed answers with probabilities, so software can act on them directly.

TypeSafe's docs list three question types.

A Choice question picks one option from a list you supply and returns a probability for every option.

A Score question rates the state against ordered levels that you define.

A Noul question answers yes or no and returns one probability between 0 and 1.

TypeSafe lists Jev at $0.042 per million input tokens, and output tokens are free.

The docs also say Jev takes text only and has a 64,000-token budget per request.

What the Jev waitlist was

The Jev waitlist was the early-access queue TypeSafe ran in launch week.

TypeSafe's launch post said it was opening early access and bringing developers off the waitlist as quickly as it could.

There was no separate form to fill in.

You created a console account and waited for approval, and that wait was the list.

TypeSafe said it had cleared roughly 140,000 sign-ups within about 36 hours.

I was one of the people waiting on launch day.

The Jev waitlist timeline, from launch to today

Access changed four times in four weeks, which is why older articles still disagree with each other.

The original queue lasted about six days, and the pause that followed lasted about five.

Some sites show these dates a day earlier, because TypeSafe posts in US time.

My first write-up of this, how I skipped the Jev waiting list, was published before the pause happened.

How to join Jev through the TypeSafe console

There is no queue form now, so joining means creating an account.

First, sign in at console.typesafe.ai.

Second, open the keys page at console.typesafe.ai/keys and create an API key.

Third, store that key in an environment variable called TYPESAFE_API_KEY.

Fourth, send a request to https://api.typesafe.ai/v1/systemone with the model set to jev-latest.

Fifth, check your balance in the console before you plan any volume.

The docs say jev-latest currently points to jev-1.13.0.

If you made an account during early access, sign in with the same email, because accounts made before the pause kept working.

There is one honest caveat on the balance.

When the queue was first removed, new accounts started with $5 of credit.

Reports from the reopening say that credit is switched off for new accounts for now.

I could not confirm the current position from TypeSafe's own docs, so read your console balance and plan from what you can see.

How to skip the Jev waitlist with no TypeSafe account

The same Jev 1.13 model is served through other doors, and none of them ever had a queue.

The fastest of the four is the OpenCode Zen free model.

You send a POST request to https://opencode.ai/zen/v1/systemone with the model set to jev-1.13-free.

I ran that on 11 October 2026 with no key in the request.

My question was whether a customer was asking for a refund, and the answer came back as 0.99 from 286 input tokens.

OpenCode has described that free model as a limited-time offer, so treat it as a trial lane.

The playground at jevplayground.com is independent and is not TypeSafe's own site.

TypeSafe's official playground is inside the console and needs a login.

My calls through OpenRouter worked out at roughly $0.00002 each.

Vercel's promotional pricing from launch month has ended, so that door now charges the normal rate.

The longer walkthrough of the free routes is in my post on the free Jev API key.

How to use Jev once you are past the waitlist

Every route takes the same three ingredients.

The first is the model name, which changes by route.

The second is the state, which is the text you want judged.

The third is your set of questions, each with a type and plain-English instructions.

TypeSafe's docs say all three question types can be mixed in one call.

They also say every question is evaluated in parallel against the same state, so adding questions barely changes the response time.

The docs advise keeping each question narrow.

If a judgement weighs several factors, you ask about each factor separately and combine the answers in your own code.

For Choice and Score questions you also get a confidence value.

The answer tells you what to do, and the confidence tells you whether to act on it.

Is there a Jev waitlist for a new tier, a preview or enterprise?

I looked for every other queue I could think of, and I did not find one.

TypeSafe's models page lists one current model, which is Jev 1.13.

The page has a jev-preview alias, and the docs say there is no preview build available right now.

So I cannot verify a waitlist for a newer or faster Jev.

For higher rate limits, the docs say custom and enterprise plans are available through TypeSafe's sales team.

That is a sales conversation and not a waiting list.

If the name "Jev Ultrafast" brought you here, I explain what it is in the Jev Ultrafast GitHub post.

Limits, gotchas and what I could not verify

Access is easy now, so the limits matter more than the door.

The docs list rate limits of 100,000 tokens per second and 80 requests per second.

The docs warn that those limits are adjusting dynamically and can change without notice.

A request over either limit returns a 429 response.

The state plus the longest question has to fit in 32,000 tokens.

English is the primary training language, and the docs tell you to test other languages yourself.

TypeSafe has paused sign-ups once already, so I would create an account now and keep one gateway key as a spare.

TypeSafe's status page showed all services online on 11 October 2026.

There are also things I could not check.

I could not load the console sign-up screen with a script, because it sits behind a browser security check.

I am relying on TypeSafe's own X post and its docs for the sign-up step.

TypeSafe's homepage still describes Jev as being in early access, and I found no statement about full general availability.

If you would rather not depend on anyone's servers, I tested the open-source options in run Jev locally.

What to do if Jev sign-ups pause again

TypeSafe's docs still say it is serving a very large volume of demand, so I build as if the door could close again.

The fix is a short fallback order that lives in your config.

Your own TypeSafe key goes first, because it is the official source and it lets you pin a version.

A gateway key such as OpenRouter goes second, because it serves the same model if the direct route is rate limited.

The OpenCode Zen free model goes third, because it keeps test jobs running at no cost while you fix the first two.

A human goes last, because anything under your confidence line should never be forced through.

Keep the endpoint and the model name in one config value, so switching routes is a one-line change.

Add a retry with backoff as well, because TypeSafe's own SDKs do that by default and a raw HTTP call does not.

Pin jev-1.13.0 once you have tuned your thresholds, because an alias such as jev-latest moves when a new release ships.

A real first job to run once you are in

Start with a triage question on an inbox you already own.

Write one Choice question with four options, which are sales, support, billing and spam.

Run ten real emails through the playground first and read the probabilities for every option.

Move the same question to the free Zen model and call it from a script.

Set a confidence line, and send everything under it to a person.

When that has run well for a week, move the call to your own TypeSafe key.

That whole path needs no approval from anyone, which is the opposite of how launch week felt.

My visual walkthrough of the free doors is on agentos.guide as 3 free ways to use Jev AI.

Why Jev has no score on this leaderboard

This is the question I get on this site in particular.

Jev does not ship a writing model, so there is nothing for this benchmark to score.

GoldieBench gives every model the same one-shot prompt to build a game, a page, a simulation or a visual, and then scores the rendered result from 0 to 10.

Jev cannot attempt that, because it does not write text or code at all.

What the board can tell you is which builders Jev should be choosing between.

A decision model is only as useful as the lanes it routes to.

The models worth pairing with Jev, from real scores

Once you are past the Jev waitlist, the next job is filling the lanes behind it.

These averages match the live model pages on 11 October 2026.

You write one Choice question with those lanes as the options.

Jev makes the call, and only the job that needs building reaches a builder.

Set a confidence line, and send anything under it up to the strong lane or to yourself.

That is how a cheap decision model protects your spend on the expensive models.

Also On Our Network

• Agent Operatorsthe operator's 5-step Jev setup, with the commands
• AI Income Deskhow a business uses Jev now the queue has gone
• AI Tool Verdictevery Jev access route scored, with a verdict
• agentos.guidethe 60-second Jev quick start for an Agent OS
• agentos.guideJulian's visual guide to the free Jev doors

The queue is gone, the console is open, and four other doors serve the same model if it ever closes again.

That is everything I know about the Jev waitlist.