Jev AI: TypeSafe’s model for typed decisions.
Jev is not a chatbot. Send it application state and it returns a choice, a score, or a yes/no probability your code can branch on. The hosted API is public as of 21 September 2026 — sign in, create a key, $0.042 per million input tokens, output free.
"The customer has been billed twice and wants a refund today."
A question in.
A typed Jev judgment out.
Jev’s value is not another chat window. It is a fast, typed branch your application can consume. Run the ticket below on the hosted Jev model — 10,000 input tokens without signup, 100,000 when you sign in.
Request preview
Your judgment
Model judgment, not a guarantee. Review the result before using it to trigger an action.
Three Jev output types cover
the fuzzy if in your system.
Route
Pick one value from a known set. Great for support topics, model routing, and intent classification.
→ scoreRank
Place input on an ordered spectrum. Use it for urgency, lead quality, risk, or review priority.
→ noulGate
Get a calibrated probability for a yes/no question before your software takes an action.
→Put the Jev decision layer
between intent and action.
Jev is most interesting when a workflow already has an LLM, an API, or an agent, but needs a small reliable branch before the next step.
See all Jev use cases