Jev
A model for structured decisions: choose an option, check a condition or evaluate text against a scale.
- Provider
- TypeSafe AI
- Version reviewed
jev-1.13.0- Guide reviewed
- Oct 8, 2026
How TypeSafe’s Jev helps your app
Jev is TypeSafe’s model for decisions inside software. Your app sends a message and a question, gets a structured answer, and uses it to route requests, flag urgent issues or categorize feedback.
Send a customer’s request to the right team without sorting it by hand.
Ready to play
Support chat
I was charged twice for this month. Can you refund the duplicate payment?
support-server Simulated logs
$ tail -f /var/log/support.log
Listening for incoming messages…
View API request and answer
API requestPOST /v1/systemone
{
"model": "jev-1.13.0",
"state": "I was charged twice for this month. Can you refund the duplicate payment?",
"questions": {
"department": {
"type": "choice",
"instructions": "Which team should handle this request?",
"criteria": {
"billing": "Payments, invoices and refunds",
"technical": "Bugs, outages and integrations",
"sales": "Plans, pricing and new accounts"
}
}
}
}Jev’s answeranswers
{
"department": {
"type": "choice",
"choice": "billing",
"probabilities": {
"billing": 0.94,
"technical": 0.04,
"sales": 0.02
},
"confidence": 0.91
}
}How it fits into your app
- Send the context. The
statecan be a message, a JSON object or an array of text values. - Define the question. Choice picks from your options. Noul returns a yes/no probability. Score evaluates text against your ordered rubric.
- Use the answer in code. Your app reads the structured result and decides whether to act, ask for clarification or send it for review.
Published pricing and limits
- Input tokens
- $0.042 per million
- Output tokens
- Free
- Total request context
- 64,000 tokens: state and all questions
- State + longest question
- 32,000 tokens
- Input format
- Text only; no image, audio or video input
Checked Oct 8, 2026 against TypeSafe’s model documentation ↗.
Calculator coverage
ByeTokens does not yet calculate Jev workloads. The calculator’s chat assumptions do not capture a shared state, multiple typed questions or both context limits. Use measured input-token usage and the published rate to plan this part of your app.
View TypeSafe’s provider overview