How to Get a Jev API Key (4 Ways, Tested)

Jev launched as early access with a waitlist on September 15, 2026. TypeSafe opened signups to everyone on September 20, then paused new signups on September 22 because of demand; existing accounts keep working. If you do not have a TypeSafe account yet, you currently cannot make one. Three platforms that resell model access listed Jev within four days of launch, and their keys work without a TypeSafe account. See Jev waitlist and signup pause.

The four routes at a glance

Route	Where the key comes from	Model name	What it takes to start	Catch
TypeSafe (official)	TypeSafe console, API keys page	jev-1.13.0 / jev-latest	An existing TypeSafe account	New signups paused since Sep 22, 2026
Vercel AI Gateway	Vercel dashboard, AI Gateway	typesafe-ai/jev	A Vercel account with a card on file	Without a card, calls fail with customer_verification_required; the free Jev promotion ended Sep 25, 2026
OpenRouter	openrouter.ai, Keys	typesafe/jev-1.13 or ~typesafe/jev-latest	An OpenRouter account with credits	No free tier for Jev; empty balance returns HTTP 402
Cloudflare	Cloudflare dashboard, API token with Workers AI Read access	typesafe/jev	A Cloudflare account with AI Gateway credits	Billed from prepaid credits; price only shown in the dashboard; not yet tested by us

If you just want to try Jev today, OpenRouter is the fastest route: add a few dollars of credit and you can call it immediately. If you plan to build on Jev, get a TypeSafe account once signups reopen, since the official API has the largest context window (64k tokens versus 32,000 on the other routes) and version pinning. The official SDKs themselves now also work against OpenRouter and Vercel by changing the base URL.

Route 1: TypeSafe’s own console

1. Sign in at console.typesafe.ai. New signups have been paused since September 22, 2026; existing accounts can sign in as usual.
2. If you signed up before the pause but were never let in from the waitlist, check the console; TypeSafe has not said how those accounts are handled.
3. Open the API keys page in the console settings and create a key. TypeSafe’s quick start also points to a browser Playground in the console, where you can paste a state and try questions without code.
4. Export it so the SDKs can find it:

export TYPESAFE_API_KEY="paste-your-key-here"

5. Make a first call with curl or an SDK; Your first Jev call has copy-paste versions.

Official pricing is $0.042 per million input tokens with free output. See Jev pricing.

Route 2: Vercel AI Gateway

1. In the Vercel dashboard, open AI Gateway and create an API key.
2. Add a payment card to the team. This is the step people miss: until a card is on file, requests to Jev fail with a customer_verification_required error. See customer_verification_required.
3. Call Jev with the AI SDK (version 7 or later) using experimental_evaluate and the model name typesafe-ai/jev. Since September 21, 2026 Vercel also offers a TypeSafe-compatible API at https://ai-gateway.vercel.sh/typesafe, so the official TypeSafe SDKs work by changing the base URL. There is still no OpenAI-compatible endpoint for Jev.

Vercel gives every team $5 of AI Gateway credit per month on the free tier, with lower per-model rate limits. In our testing the free tier let a handful of Jev questions through and then throttled for a while; 131 questions took about an hour and a half. Buying credits moves the team to the paid tier, which lifts the limits, but Vercel’s docs note that the monthly free credit then stops.

Jev was free on Vercel during a launch promotion that ended on September 25, 2026. Vercel now lists it at $0.042 per million input tokens with free output, the same as TypeSafe. Full walkthrough: Jev on Vercel AI Gateway.

Route 3: OpenRouter

1. Create an OpenRouter account and an API key.
2. Add credits. OpenRouter lists Jev at $0.042 per million input tokens and $0 for output, the same rates as TypeSafe. With no balance you get HTTP 402, “Insufficient credits”.
3. Send requests to OpenRouter’s Decisions endpoint, POST https://openrouter.ai/api/alpha/decisions, with the model typesafe/jev-1.13. OpenRouter also accepts TypeSafe’s own request format at POST https://openrouter.ai/api/v1/systemone, so the official SDKs work with the base URL https://openrouter.ai/api. The regular chat completions endpoint does not serve Jev, and the default /api/v1/models list leaves it out; add ?output_modalities=decisions to see it.

Details and a working request: Jev on OpenRouter.

Route 4: Cloudflare Workers AI

Cloudflare added Jev to its model catalog as typesafe/jev, a third-party model served through Cloudflare’s AI Gateway, with a 32,000-token context window. You call it from a Worker with env.AI.run('typesafe/jev', {...}) or through the REST API with a Cloudflare API token that has Workers AI Read permission. It is billed from prepaid AI Gateway credits rather than Workers AI Neurons, and the price is shown only inside the Cloudflare dashboard. We have not run a paid call yet; see Jev on Cloudflare Workers AI for what the documentation says.

Keep the key out of the browser

A Jev key is billed per token. Call Jev from your server, a serverless function or a Worker, not from front-end code. The official JavaScript SDK refuses to run in a browser unless you set an option named dangerouslyAllowBrowser, which exists for a reason. Never commit the key to a repository; load it from an environment variable.

Troubleshooting a new key

You see	It means	Page
HTTP 401, “Cannot authenticate with the server”	The key is wrong or revoked	401
HTTP 403, “Must supply an API key!”	The request had no Authorization header	403
customer_verification_required	Vercel team has no card on file	Vercel error
HTTP 402, “Insufficient credits”	OpenRouter balance is empty	402
HTTP 400, “Unknown model”	Wrong model name for that route	400