Jev 1.13
typesafe/jev-1.13
Typed decisions for production classification and routing.
See which team should handle a support ticket, how urgent it is, and whether it needs human review. Try a duplicate-charge example, then test your own text or JSON.
typesafe/jev-1.13
Typed decisions for production classification and routing.
upstage/solar-decide
Decision model for classification and scoring tasks.
upstage/solar-decide-flash
Low latency variant for common structured decisions.
Use one model for typed answers across your app, API, and batch workflows.
Use the same typed questions in your app with one server-side API key. Select models, compare answers, and scale with Batch.
curl https://ai-jev.pro/api/v1/systemone \
-H "Authorization: Bearer $JEV_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"typesafe/jev-1.13","state":"...","questions":{"urgent":{"type":"noul","instructions":"Is this urgent?"}}}'Read API docs →For trying a few models
Never expires. Shared across Playground, API, and Batch.
For production workflows
Never expires. Shared across Playground, API, and Batch.
For higher volume
Never expires. Shared across Playground, API, and Batch.
Input tokens are charged at the selected model's rate. Output tokens are free.
Choose an answer space once, then route, review, or score each request with structured output.
Check whether a message needs a person to take a closer look.
Send a ticket or lead to the team or category you define.
Score urgency, fit, or quality against your own labels.
Jev turns a shared text or JSON state into typed Yes/No, Choice, and Score answers for workflows such as support triage, moderation, routing, and lead qualification.
Choose from 18 OpenRouter models. Model IDs and input-token credit rates are listed in the model directory.
Yes. A request can include up to 8 named questions. Span models accept Noul yes/no questions only.
Input tokens are charged at the selected model rate. Output tokens are free. New Jev accounts receive 100 welcome credits; top-up packs do not expire.
Vision models accept up to four PNG, JPEG, or WebP images, up to 4 MiB each and 8 MiB total. Check each model's image support before running.
Completed run content is not saved to an account history. Saved configurations and credit or billing records are stored. The model provider's handling terms apply; see the Privacy Policy.
Use your Jev API key to run a typed decision from your server.
Read quickstart →BATCHApply the same questions to a collection of rows and export results.
Open Batch →USE CASERoute tickets, set priority, and flag cases for human review.
Explore use case →