Questions? hello@viberation.devGet supportBlogDocsChangelog
Get started

Returns a typed decision and a confidence, not a paragraph.

Jev is TypeSafe AI's first "System One" model. You give it some state and a question whose possible answers you have already defined, and it returns one of them as a type your code can use — a choice, a score or a true/false — along with how confident it is. It does not write prose or code, and it cannot read images or audio, so it sits alongside a model like Claude rather than replacing one. TypeSafe are clear about the catch, and it is worth repeating: a typed answer is guaranteed to be the right shape, not the right decision, so check it against real outcomes before you let it route anything that matters.

Pairs well with

Key info

Pricing
Paid
Category
Models
Best for
Expert
Made by
TypeSafe AI
Answers with
A choice, a score or a true/false, plus a probability
Takes
Text, JSON and arrays
Reach it through
Vercel AI Gateway, using the AI SDK's experimental eval API
Availability
Early access
Not for
Writing prose or code, or reading images and audio

More in Models

  • OpenAI's model family, behind ChatGPT and Codex.

    Models
  • Anthropic's model family, strong at long-context coding work.

    Models
  • Google DeepMind's multimodal model family.

    Models