Reference · independent

Jev API

The Jev API is the community name for TypeSafe AI’s System One HTTP interface — api.typesafe.ai/v1/systemone. You send state plus typed questions and receive constrained answers (Choice, Score, or Noul), not free-form text.

This page covers the endpoint, model naming, official list pricing, and SDKs. For the conceptual explainer, read what is Jev AI. Figures below are vendor-published and attributed — this directory does not run its own benchmarks.

Endpoint and authentication

Production traffic goes to TypeSafe’s hosted System One path. This site does not proxy keys or rewrite the contract — use first-party docs when wiring production.

HTTP endpointhttps://api.typesafe.ai/v1/systemone

AuthAPI key from TypeSafe (commonly TYPESAFE_API_KEY in SDK examples). Keys are not issued by jevaitools.com.

AccessAsk about waitlists, limits, and previews on TypeSafe Discord.

Official docsdocs.typesafe.ai/api

Jev model

Searchers looking up “jev model” usually mean TypeSafe’s System One decision model — the thing behind the Jev API — not a chat LLM. Community and gateway listings sometimes label a hosted route as jev-latest; treat that as a deployment label, not a separate open-weight checkpoint from this directory.

The model returns constrained answers for questions you declare up front: pick from your options (Choice), a level on your scale (Score), or a yes/no gate (Noul). It is not designed to write prose, summaries, or chat replies.

Parameter counts, architecture papers, and weight releases are not documented in the vendor materials reviewed here. For product framing and when the approach fits, see what is Jev AI.

Jev pricing

Pricing below is TypeSafe’s published list rate for System One / Jev API usage. Quote it as vendor-published; confirm on the official site before you set budgets.

List priceOfficial list price: $0.042 / 1M input tokens; output tokens free (source: typesafe.ai, as of 2026-09).

Published latencyOfficial end-to-end latency range: ~70–500ms; many calls land near ~100ms from US West Coast (source: typesafe.ai / TypeSafe public materials, 2026-09).

Verify before you rely on either figure: typesafe.ai · docs.typesafe.ai

Official SDKs

Prefer first-party clients when you need the same question helpers and response shapes documented for System One.

TypeSafe JavaScript SDK

@typesafe-ai/sdk · Node 20+

npm install @typesafe-ai/sdk

TypeSafe Python SDK

Official Python client for Choice / Score / Noul via system_one

uv add typesafe-sdk

systemOne.ts

import { TypeSafe } from '@typesafe-ai/sdk'

const client = new TypeSafe({ apiKey: process.env.TYPESAFE_API_KEY })

const result = await client.systemOne({
  state: 'User asks to cancel subscription',
  questions: [
    {
      id: 'intent',
      type: 'Choice',
      question: 'What should we do?',
      options: ['CANCEL', 'RETAIN', 'ESCALATE'],
    },
  ],
})

console.log(result.answers.intent)

Where to go next

what is Jev AI

Conceptual explainer: typed decisions, Choice / Score / Noul, and when the approach fits.

try Jev in the browser

Live playground call with measured round-trip latency for that request.

Jev vs Laya

Hosted API versus open-weight checkpoints — every figure attributed.

Official category

First-party SDKs and adapters listed in the directory catalog.

Frequently asked questions

What is the Jev API endpoint?

The hosted System One HTTP endpoint is https://api.typesafe.ai/v1/systemone. Community materials often call this the Jev API. This directory quotes the path as published by TypeSafe and does not operate the service.

How do I get a Jev API key?

Keys are issued by TypeSafe, not by this site. Ask about access, limits, and previews on the TypeSafe Discord (https://discord.gg/typesafe), then call the endpoint or use an official SDK once you have a key.

How much does the Jev API cost?

Official list pricing as published by TypeSafe in September 2026 is $0.042 per 1M input tokens with output tokens free. Verify current rates on https://typesafe.ai before budgeting — this page does not run its own pricing tests.

What is the Jev model?

In community usage, “Jev model” usually means TypeSafe’s System One decision model (sometimes referenced as jev-latest on third-party gateways). It answers typed Choice, Score, and Noul questions instead of generating free text. Architecture details and parameter counts are not published in the vendor materials reviewed here.

Which SDKs call the Jev API?

The official JavaScript/TypeScript client is @typesafe-ai/sdk. The official Python client is published as typesafe-sdk. Both target the same System One request shape. Prefer the official packages over unofficial wrappers when you need calibrated answers and documented latency/pricing.

Can I try the Jev API without writing code?

Yes. This site’s browser playground proxies a live decision call so you can paste state, ask a typed question, and see the structured answer with measured round-trip latency for that request.