Skip to content

Jev

TypeSafe AI's System One model: returns typed decisions with probabilities instead of text.

By TypeSafe AIPay as you goSince 2026LLMs

Jev is a model from TypeSafe AI, a San Francisco company, that returns typed decisions instead of generated text. You send it some content and a set of typed questions, and it answers each one with a yes/no probability, a choice from your options, or a score on your scale. TypeSafe announced it on 15 September 2026 and bills input tokens at $0.042 per million, with output tokens free.

Facts checked on against the sources listed at the end of this page.

About Jev

Jev is the first model in what TypeSafe calls the System One class: models built to make fast, structured decisions that software can use directly, rather than to write text. It is meant for classification, routing, scoring, extraction and branching inside a workflow, where an application needs an answer it can act on.

You call the API with the content to evaluate (a string, object or array) and a map of typed questions. Jev returns one answer per question. A yes/no question returns a probability between 0 and 1. A choice question returns the selected option and a probability for each option. A score question returns a rating along a rubric you define.

TypeSafe says end-to-end latency is 70 to 500 milliseconds. It also says the model is trained with a method it calls Reinforcement Learning for Calibrated Decisions, so the probabilities it returns reflect its confidence. Jev does not generate text strings: TypeSafe describes this as the trade-off that makes the typed output possible.

A good fit for

  • Classifying or routing high volumes of text, such as support tickets, where cost per call matters
  • Moderation, triage and scoring steps that need a probability your code can threshold
  • Replacing a chat model that you only prompt to return a label or JSON
  • Workflows where response time of a few hundred milliseconds matters

Not a good fit for

  • Writing, summarizing or chat: Jev can't generate text
  • Open-ended reasoning or agent work that needs free-form output
  • Self-hosting: it is a proprietary hosted model
  • Teams that need long-term production history: it was announced on 15 September 2026

Main features

Typed answers
Three question types: yes/no, choice from a list you define, and score on a rubric you define.
Probabilities with every answer
Each answer includes probabilities, so code can branch on confidence or send low-confidence cases to a person.
Low latency
TypeSafe reports 70 to 500 milliseconds end to end.
Free output tokens
Only input tokens are billed, at $0.042 per million.
Single REST endpoint
POST to https://api.typesafe.ai/v1/systemone with a bearer token and the model name jev-latest.
Vercel AI Gateway
Also available through Vercel's AI Gateway, so it can be called alongside other models.

Jev pricing

TypeSafe prices Jev at $0.042 per million input tokens, and output tokens are free. There are no seats or plans listed.

PlanPriceWhat you get
Jev$0.042 per million input tokensOutput tokens are free.

Prices in USD, checked on 29 September 2026. Current prices on typesafe.ai.

Getting started

  1. Create an account in the TypeSafe console at console.typesafe.ai and generate an API key.

  2. Send requests to the REST endpoint with your key as a bearer token. Set model to jev-latest, pass the content as state and your typed questions as questions.

    POST https://api.typesafe.ai/v1/systemone
    Authorization: Bearer $TYPESAFE_API_KEY
  3. Read the answers map in the response. Each question returns its typed answer with probabilities, and usage reports input and output tokens.

Questions about Jev

What is Jev?

Jev is TypeSafe AI's first System One model. It returns typed decisions with probabilities, such as a yes/no probability or a choice from a list, instead of generated text.

How much does Jev cost?

TypeSafe lists $0.042 per million input tokens. Output tokens are free.

Can Jev replace ChatGPT or Claude?

No. Jev can't write or chat. It suits classification, routing, scoring and extraction steps that you would otherwise prompt a chat model to do.

Is Jev open source?

No. Jev is a proprietary model available through the TypeSafe API and Vercel AI Gateway.

How fast is Jev?

TypeSafe reports end-to-end latency of 70 to 500 milliseconds. That is the company's own figure.

Sources

Jev: pricing, features and alternatives (2026) | ScratchDB