Provider · TypeSafe AI
Jev 1.13
The hosted Jev decision model for classification, scoring and routing with typed, structured answers.
API model ID
typesafe/jev-1.13ChoiceScoreNoul
Site usage rate
600 credits / million input tokens
MODEL CATALOG
Find the model for your decision workflow.
Compare the models you can call through the Jev API and playground. Each card shows its exact API model ID, supported response types, billing rate and a route to try it.
9 supported model IDsChoice · score · yes/no decisionsConversation behavior evaluation
Structured decision models
Use one state with typed questions that return choices, scores or probabilities.
Decision
Provider · Liquid AI · Vercel AI Gateway
Liquid d1
Liquid AI's decision model, connected through Vercel AI Gateway with the same TypeSafe request and answer schema as Jev.
API model ID
liquid/d1ChoiceScoreNoul
Site usage rate
No token-based charge · 1 credit minimum per successful request
Provider · Upstage
Solar Decide
A hosted model for typed decisions, including classification, ordered scoring and yes/no behavior checks.
API model ID
upstage/solar-decideChoiceScoreNoul
Site usage rate
720 credits / million input tokens
Provider · Jared Palmer
Kev 4B
A compact hosted decision model that returns structured choices, scores or yes/no answers.
API model ID
jaredpalmer/kev-4bChoiceScoreNoul
Site usage rate
600 credits / million input tokens
Provider · Laya
Laya English
An English language route for structured classification, scoring and yes/no decisions.
API model ID
laya-englishEnglishChoiceScoreNoul
Site usage rate
600 credits / million input tokens
Provider · Laya
Laya Multilingual
A multilingual route for structured decisions across supported languages.
API model ID
laya-multilingualMultilingualChoiceScoreNoul
Site usage rate
600 credits / million input tokens
Provider · Laya
Laya Auto
Automatically routes supported language input to an English or multilingual Laya model.
API model ID
laya-autoAutomatic routingChoiceScoreNoul
Site usage rate
600 credits / million input tokens
Conversation behavior models
These models evaluate whether a defined behavior appears in an assistant response. They accept Noul yes/no questions.
Provider · Respan
Span-01
A conversation evaluation model for checking defined behaviors in assistant responses.
API model ID
respan/span-01NoulBehavior evaluation
Site usage rate
300 credits / million input tokens
Provider · Respan
Span-01 Lite
A lightweight conversation evaluation route for yes/no behavior checks.
API model ID
respan/span-01-liteNoulBehavior evaluation
Site usage rate
No token-based charge · 1 credit minimum per successful request
$1 buys 10,000 credits. Input usage is rounded up per request, with a 1-credit minimum; output tokens are not charged. Rates shown here are the site's credit rates, not the provider's list prices. Liquid d1 and Span-01 Lite currently have no token-based charge and use the per-request minimum.