4 Ways Past The Jev Waiting List (All Tested)

Q: Which route gives Jev access without any waitlist?

OpenRouter (typesafe/jev-1.13), Vercel AI Gateway (typesafe-ai/jev) and Cloudflare Workers AI (typesafe/jev) never required approval. Direct access through TypeSafe is also open now.

If you are stuck on the Jev waiting list, the good news is that it no longer exists — and even while it did, there were three other ways to call the exact same model.

TypeSafe removed the queue on 21 September 2026, six days after launch.

But I want to spend this post on something more useful than that one fact, because the same situation will happen with the next hot model launch.

There are four routes to Jev 1.13.

They all return identical answers, and they differ on what you get around the model — SDKs, version pinning, billing and where the code runs.

I have used all four, so here is the honest comparison and which one I would pick for each situation.

Compared at a glance

Route	Model ID	Waitlist	Free credit	Official SDKs	Version pinning	Best for
TypeSafe direct	`jev-1.13`	Gone since 21 Sep 2026	$5 (~120M tokens)	Yes, Python + JS	Yes	Production
OpenRouter	`typesafe/jev-1.13`	Never had one	No, prepaid credits	No, generic client	No	Multi-model stacks
Vercel AI Gateway	`typesafe-ai/jev`	Never had one	No, card on team	AI SDK 7 `evaluate`	No	Apps already on Vercel
Cloudflare Workers AI	`typesafe/jev`	Never had one	No, CF account	No, Workers binding	No	Edge decisions

Every one of those serves Jev 1.13.

The model does not change, only the plumbing around it.

Route 1: Straight through TypeSafe (what I'd pick now)

This is the one that used to have the queue, and it is now the one I would recommend to almost everyone.

You register at console.typesafe.ai, confirm your email and create an API key.

The account comes with $5 of credit, which is about 120 million input tokens at Jev's rate.

Set the key as TYPESAFE_API_KEY in your environment and install the official SDK for your language.

If you are on LangChain already, the langchain-typesafe package picks the variable up automatically.

What you get here that you do not get anywhere else matters more than it sounds.

You get version pinning, so you can lock to jev-1.13.0 and stop a silent model upgrade shifting your thresholds overnight.

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

You get direct status page access, which is worth something given the API went down during launch week.

And you get a larger documented request budget than the gateways give you.

I walked through the whole setup in the how to use Jev guide.

🔥 Want the exact setup I used to get these results? Inside the AI Profit Boardroom, I've got a full AI model and API section with step-by-step video tutorials on exactly this kind of stack. Plus weekly coaching calls + 2,800+ members building real automations. → Get access here

Route 2: OpenRouter

OpenRouter never had a waiting list for Jev, and it was the first place a lot of people got their hands on it.

The model ID is typesafe/jev-1.13, and you need prepaid credits on your OpenRouter account.

The obvious win is unified billing.

If you already route five models through OpenRouter, adding a sixth is one string change and it lands on the same invoice.

The trade is that you are using a generic client rather than TypeSafe's SDK, and you cannot pin a version.

For a prototype that is completely fine.

For something where a threshold has been tuned over weeks, it is a risk you are taking without noticing.

Route 3: Vercel AI Gateway

Vercel had Jev live as typesafe-ai/jev from 16 September, one day after launch and with no approval step.

You call it from AI SDK 7's experimental evaluate method with a gateway key.

All you need is a card on the team.

This is genuinely the fastest route if your app already lives on Vercel — it is two lines of config rather than a new account and a new key to rotate.

It is also where the adoption numbers came from.

Vercel said Jev became the fastest-adopted model in AI Gateway history, used by nearly 13% of paid teams by hour 24, more than twice the share of any previous model launch.

Same caveats as OpenRouter — no official SDK, no version pinning.

Route 4: Cloudflare Workers AI

The model ID here is typesafe/jev, and all you need is a Cloudflare account.

This is the interesting one for a specific shape of problem.

If your decision needs to happen at the edge — routing a request before it ever reaches your origin, or gating something in a Worker — running Jev there removes a whole round trip.

Given the model already answers in 70ms to 500ms, shaving the network hop is a real percentage of total latency.

It is a narrower use case than the others, but when it fits, nothing else comes close.

What you are actually calling, whichever route you pick

Worth covering, because plenty of people joined the waiting list without checking what Jev does.

Jev is a decision model, not a language model.

You pass it unstructured state plus a set of questions, and it returns typed answers with calibrated probabilities.

It never writes prose, and the output is meant for your code rather than a human reader.

There are three question types and that is the whole API surface.

Choice picks one option from a list you supply and returns a probability for every option plus an overall confidence number.

Score rates the state against ordered levels you define and returns a continuous score plus a distribution across those levels.

Noul gives a straight yes or no with a single probability that the answer is true.

Choice caps at 255 options per query, so picking from thousands of items means scoring in batches first and running the finalists through one Choice question.

There is a deeper breakdown in the Jev AI model post and Jev architecture.

The cost comparison that makes all this worth doing

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

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

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

That pairing is the reason the waiting list filled up in the first place.

A decision cheap enough and fast enough to run inside a hot loop changes the architecture of your agent, not just its bill.

Detailed numbers are in the Jev pricing post.

🔥 Building an AI stack that has to stay cheap at scale? The AI Profit Boardroom covers model routing, fallbacks and cost control end to end — the same systems I run in my agency. Weekly coaching calls, 2,800+ members. → Join the Boardroom

My honest pick for each situation

If you are shipping something customers touch, go direct through TypeSafe.

Version pinning alone justifies it, and now that the queue is gone there is no reason not to.

If you already have a multi-model stack on OpenRouter, add Jev there first and move it later if it sticks.

If your app is on Vercel and you want to try Jev this afternoon, the gateway is the shortest path in existence.

If the decision needs to happen before the request reaches your origin, use Workers AI.

And if you are just curious, take the $5 credit and spend an hour on it — that is 120 million tokens of experimentation for nothing.

What to watch out for on every route

Rate limits are still being adjusted dynamically against available GPU capacity, and TypeSafe has not published fixed numbers.

Build retries with exponential backoff into your client from the start.

TypeSafe has not said whether dropping the queue means full general availability or an open beta, so do not assume stability guarantees you have not been given.

There are no published credit-expiry terms on the $5.

And on accuracy, TypeSafe's own disclosed numbers put Jev at 67.8% aggregate against 74.1% for a comparable LLM, with a self-published JevBench score of 75.3.

Run your own evaluation set against your own production data before you commit.

Demo prompts flatter every model ever released.

Jev waiting list FAQ

Do I still need to get past the Jev waiting list?

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

Which route gives Jev access without any waitlist?

OpenRouter (typesafe/jev-1.13), Vercel AI Gateway (typesafe-ai/jev) and Cloudflare Workers AI (typesafe/jev) never required approval. Direct access through TypeSafe is also open now.

Is the model different on OpenRouter or Vercel?

No. All four routes serve the same Jev 1.13 model and return identical answers. What differs is the SDK, version pinning, billing and where your code runs.

Why go direct if the gateways work fine?

Direct access gives you the official Python and JavaScript SDKs, version pinning to a specific build like jev-1.13.0, the status page and a larger documented request budget. Gateways give you none of those.

How much testing does the $5 free credit cover?

Roughly 120 million input tokens at $0.042 per million. Output tokens are not charged, so the credit stretches a long way for a decision model.

Can Jev write text if I ask it to?

No, and that is deliberate. The API returns typed decisions only — choices, scores and yes/no answers with calibrated probabilities. Keep a generator model for prose and let Jev handle the routing and verification around it.

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.