Jev PlaygroundIndependent playground
STRUCTURED DECISIONS, REAL PROBABILITIES

Jev Playground

Run the real TypeSafe Jev model online.
Test structured AI decisions — no signup required.

Real JevTypeSafe AI modelNo signup
TRY AN EXAMPLE
Evaluation inputs
INPUT
118 / 8,000
Choices4 of 10

Free to try · Usage limits apply

OUTPUTTypeSafe Jev
Your Jev decision will appear here.

Your Jev decision will appear here.

Give Jev a state and a typed question.
Run it to see the decision and probabilities.

StateJevDecision
Real model execution. No simulated answers.

No TypeSafe account required. Choose an example, edit the inputs, then run the real model.

01 / THE MODEL

What is Jev?

Jev is a System One model from TypeSafe AI. It is designed for fast, structured decisions rather than open-ended chat.

Give it context and a typed question. Get a bounded answer with probabilities your software can use.

Read the guide: what Jev is and how it works

02 / DECISION TYPES

Three ways to use Jev

The same context. The right kind of answer.

Boolean

Estimate how likely something is to be true.

“Was a refund issued?”
Returns P(true)

Choice

Select one option from a defined set.

“Which team should handle this?”
Returns a choice + probabilities

Score

Evaluate a situation against an ordered rubric.

“How urgent is this incident?”
Returns a score + probabilities
03 / A DIFFERENT OUTPUT

Decisions, rather than a conversation.

General-purpose LLMs often produce open-ended text. Jev is designed for bounded, typed decisions with probabilities. The right model depends on the task.

General-purpose LLMs

Drafting, explaining, open-ended chat

Generated text
TypeSafe Jev

Routing, verifying, scoring

Typed decision + probability
04 / IN YOUR WORKFLOW

Small decisions. Useful building blocks.

01Support routing02Verification03Risk classification04Agent next-action selection05Workflow gating06Urgency scoring
05 / FOR DEVELOPERS

From playground
to your application.

Use the AI SDK’s experimental evaluation API through Vercel AI Gateway. Keep your Gateway key on the server.

Read the evaluation API docs
evaluate.tsAI SDK · Server-side
import { createGateway, experimental_evaluate as evaluate } from 'ai';

const gateway = createGateway({
  apiKey: process.env.AI_GATEWAY_API_KEY,
});

const result = await evaluate({
  model: gateway.evaluationModel('typesafe-ai/jev'),
  state: 'The support agent issued a full refund.',
  questions: {
    refunded: { type: 'boolean', instructions: 'Was a refund issued?' },
  },
  providerOptions: { gateway: { zeroDataRetention: true } },
});

console.log(result.answers.refunded.probability);
06 / ACCESS

Free to try. No signup.

This independent playground is free to use, with usage limits to keep it available. Requests may be temporarily rate limited.

Building with the API?

Your own application uses your Vercel AI Gateway account and its usage-based pricing.

View current Jev API pricing