Jev availability guide · read October 7, 2026
Where to Run Jev
Five places serve TypeSafe’s Jev model today. The three big clouds do not. Model IDs, endpoints and what to do instead.
Jev is a hosted model with no published weights, so “running Jev” means choosing an endpoint. This page lists every platform that serves jev-1.13.0 as of October 7, 2026, with the exact model name and the endpoint each one uses, and says plainly where Jev is not listed so you do not go looking for it in a model catalogue that does not have it.
Where Jev is available
Same model, same request shape. What differs is the key, the bill and what sits around the call.
TypeSafe API
Available- Model
jev-latest · jev-preview · jev-1.13.0- Endpoint
- POST https://api.typesafe.ai/v1/systemone, Bearer key from the TypeSafe console
- The source: $0.042 per million input tokens, output free, 64k context, rate limits published and adjusted dynamically.
- Official SDKs for JavaScript/TypeScript and Python. Several independent guides describe console access as invite-gated at times; the gateways below need no invite.
OpenRouter
Available- Model
typesafe/jev-latest · typesafe/jev-1.13 · typesafe/jev-1.13-20260917 · typesafe/jev-router- Endpoint
- OpenRouter’s Decisions API, POST https://openrouter.ai/api/alpha/decisions, with an OpenRouter key
- One key for Jev and the generative models you pair it with. The dated 1.13 snapshot is the only public record of when the build shipped.
- A decisions endpoint, not chat completions: send state and typed questions, not messages.
Vercel AI Gateway
Available- Model
typesafe-ai/jev- Endpoint
- Point a TypeSafe client at https://ai-gateway.vercel.sh/typesafe with a gateway key, call POST /v1/evaluate over HTTP, or use experimental_evaluate in the AI SDK
- Added on 21 September 2026. Existing systemOne calls work unchanged once the base URL and key are swapped.
- Billed through AI Gateway alongside your other model calls, with its usage and observability.
Vercel changelog — AI Gateway supports TypeSafe clients and an HTTP API for Jev ↗
Cloudflare Workers AI
Available- Model
typesafe/jev- Endpoint
- The Workers AI run endpoint, from a Worker binding or POST /accounts/{account}/ai/run/typesafe/jev with a Cloudflare API token
- Runs at Cloudflare’s edge next to your Workers. No waitlist.
- Cloudflare also trains its own Jev-compatible models, Clef and Clef Flash, on the same platform with image input and open weights.
Jev AI (this site)
Available- Model
jev-latest · jev-preview · jev-1.13.0, plus Laya, Mercury Decide and Clef- Endpoint
- POST https://jev-ai.pro/api/v1/systemone with a Jev AI key; TypeSafe’s SDK works with baseURL https://jev-ai.pro/api
- A playground to try the model first, saved judges you can call by ID, batch runs and a credit balance instead of a card on file.
- Metered on input tokens with credits as the fallback; welcome credits on a new account.
Where Jev is not listed
Checked on October 7, 2026. Each cloud can still call Jev from its serverless runtime.
Amazon Bedrock
Not listed- Status
- Not a Bedrock model
- Detail
- No inference profile, VPC endpoint or Guardrails integration as of the last published check
- On AWS, Jev runs as a sidecar: Lambda or AgentCore Runtime calls TypeSafe, OpenRouter or Jev AI for classify, route, score and verify, and Bedrock stays the generation layer.
Google Vertex AI
Not listed- Status
- Not in the Model Garden
- Detail
- We found no Vertex AI listing for Jev as of October 7, 2026
- Call one of the hosted endpoints above from Cloud Run or Cloud Functions; Gemini stays the generation layer.
Azure AI Foundry
Not listed- Status
- Not in the Foundry catalogue
- Detail
- We found no Azure AI Foundry listing for Jev as of October 7, 2026
- Call one of the hosted endpoints above from Azure Functions or Container Apps; Azure OpenAI stays the generation layer.
Choosing between them
The request is the same everywhere, so pick on operations, not on the model.
| If you | Use |
|---|---|
| Have TypeSafe console access and want list price | TypeSafe API |
| Already route Claude, GPT or Gemini through one gateway and want one key and one bill | OpenRouter or Vercel AI Gateway |
| Run your code in Cloudflare Workers | Cloudflare Workers AI, or Clef if you want image input and open weights |
| Want to try the model on your own text first, save judges and pay with credits | Jev AI |
| Need the model inside your own network | Not possible with Jev; see the open alternatives |
One request, any endpoint
Only the host and the header change. The state and the typed questions stay the same.
curl --fail-with-body https://jev-ai.pro/api/v1/systemone \
-H "Authorization: Bearer $JEV_AI_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "jev-latest",
"state": "Chat: my card was charged twice this month.",
"questions": {
"team": { "type": "choice", "instructions": "Which team should handle this?",
"criteria": { "billing": null, "technical": null, "account": null } }
}
}'Swap the host for api.typesafe.ai, the gateway’s TypeSafe base URL or OpenRouter’s decisions endpoint and the body is unchanged. The Jev 1.13.0 guide has the version’s limits; the documentation has the full reference.
Where to run Jev FAQ
Is Jev available on AWS Bedrock?
No. Jev is not a Bedrock model: there is no inference profile, VPC endpoint or Guardrails integration. On AWS, call TypeSafe, OpenRouter, Vercel AI Gateway or Jev AI from Lambda or AgentCore Runtime and keep Bedrock for generation.
Is Jev on Vertex AI or Azure AI Foundry?
We found no listing for Jev in the Vertex AI Model Garden or the Azure AI Foundry catalogue as of October 7, 2026. Both clouds can call the hosted endpoints on this page from their serverless runtimes.
What is Jev’s model ID on OpenRouter?
OpenRouter’s TypeSafe page lists typesafe/jev-latest, typesafe/jev-1.13, the dated snapshot typesafe/jev-1.13-20260917 and typesafe/jev-router, all served through its Decisions API rather than chat completions.
How do I call Jev through Vercel AI Gateway?
Three ways, all added on 21 September 2026: point an existing TypeSafe client at the gateway’s TypeSafe base URL with a gateway key, POST to /v1/evaluate over HTTP, or use experimental_evaluate in the AI SDK. The model ID is typesafe-ai/jev.
Is Jev on Cloudflare Workers AI?
Yes, as typesafe/jev, callable from a Worker binding or the Workers AI REST endpoint with no waitlist. Cloudflare also offers its own Jev-compatible Clef and Clef Flash models there, which Jev AI hosts too.
Which option should I pick?
TypeSafe direct if you have console access and want list price. OpenRouter or Vercel AI Gateway if you already route other models through them and want one key and one bill. Cloudflare if your code runs in Workers. Jev AI if you want to try the model in a playground first, save judges and pay with credits.
Can I run Jev locally?
No. Jev is a hosted, proprietary model with no published weights. The open alternatives measured on the same benchmark, such as Cygnet, Winnow-12B, decider-4b and JevK5, each have a guide on this site with hardware requirements.
TypeSafe, OpenRouter, Vercel, Cloudflare, Amazon Web Services, Google Cloud and Microsoft Azure are products of their respective owners and are not affiliated with Jev AI. Availability was checked on October 7, 2026 and can change.