Jev model: official website and identity
Jev is TypeSafe AI’s first System One decision model. It evaluates typed questions against context and returns structured decisions for software. This page explains its identity and collects verified links to the TypeSafe homepage, documentation, and console for learning about the product, trying it, and accessing its API.
Official Jev website and entry points
Section titled “Official Jev website and entry points”| Your goal | Official entry point | Verification basis |
|---|---|---|
| Identify the product and developer | TypeSafe AI homepage: typesafe.ai | Introduces Jev and links directly to Docs and API console |
| Read about capabilities and integration | TypeSafe documentation: docs.typesafe.ai | Linked from the homepage; Introduction identifies Jev as TypeSafe’s first System One model |
| Sign in, get an API key, and try it | TypeSafe console: console.typesafe.ai | Linked from the homepage’s API console navigation and the official Quick start |
Entry points verified on 2026-10-19. We read the homepage navigation, documentation, and console login page; we did not sign in or verify actions behind login. Other domains or community projects claiming to be the official Jev site have unconfirmed official status unless supported by an official source. A name alone does not establish ownership.
How to start
Section titled “How to start”- Learn about Jev: read the official Introduction, or our Capabilities and Limitations.
- Try it in a browser: follow the official Playground instructions, sign in to the console, supply a state, add questions, and inspect answers. This is a structured-decision interface rather than general chat.
- Integrate it: follow the official Quick start to get a console API key, then use our Get started and API guide. Check the official console and documentation for account access, available credit, and current prices. Do not submit API keys to this site.
What Jev does and how it differs from chat models
Section titled “What Jev does and how it differs from chat models”The official System One documentation explains that Jev understands natural-language input and returns typed decisions and probabilities rather than articles, code, or reasoning explanations. It currently accepts text, JSON objects, and arrays of text, but not images, audio, or video.
The official Introduction lists three question types that can be combined in one request:
- Choice selects from supplied options and returns
choice,probabilities, andconfidence. - Score evaluates a supplied rubric and returns
score,probabilities, andconfidence. - Noul evaluates whether a statement is true and returns a
noulvalue from 0 to 1.
For example, an application can supply a ticket as state, use questions to ask about its department and urgency, and decide routing or review in code. The official Quick start uses POST https://api.typesafe.ai/v1/systemone; see the API guide for fields and code.
Constrained answer types do not guarantee correct judgments. The official System One documentation says calibration does not guarantee an individual answer is correct. Business actions still need deterministic checks and review policies; see our ticket-triage lab.
Where the names come from
Section titled “Where the names come from”The official launch blog connects Jev’s name to William Stanley Jevons and the idea that efficiency can increase demand. System One draws on fast, intuitive thinking in Kahneman’s Thinking, Fast and Slow. These are the official naming explanations.
SDKs, community integrations, and replicas in our Ecosystem have their own maintainers. Inclusion does not imply TypeSafe endorsement or that Jev model weights are open.
Continue reading
Section titled “Continue reading”- Capabilities: question types and parallel questions
- Limitations: constraints, pricing sources, and decision risks
- Get started: from the official console to a first API request
- About this site: maintainers, sourcing, and corrections