Question 1
One key and description per line: billing: Payments or refunds
LIVE JEV PLAYGROUND
Try Jev in your browser — type a question, pick a question type, and get a real Choice, Score or Noul answer in seconds. Free, no API key, no sign-up
LIVE JEV PLAYGROUND
Free to use with no API key or sign-up. We do not store your submissions.
Paste the ticket, email, log, user input, or model output that every question should evaluate.
285 / 8,000
Ask up to six typed questions about the same state. Jev answers them in one request.
Question 1
One key and description per line: billing: Payments or refunds
Question 2
Enter 2–10 levels, lowest to highest, one per line.
Question 3
Use exactly two lines beginning with true: and false:.
Try a preset
One request evaluates every question above.
Your state and questions are processed by Jev. We do not store them or include them in analytics.
Run the preset to see real Jev answers, probabilities, confidence, latency, tokens, and cost.
Choice and Noul answers at or above this confidence can be marked for automatic action.
After testing a Choice question, a Score scale, or a Noul check, try one of the free Jev AI tools for a ready-made classification, spam, sentiment, or email workflow.
DECISIONS, NOT CHAT
The Jev Playground evaluates text or structured state against questions you define. Jev returns typed decisions and probabilities for software to use, rather than composing an open-ended response.
CHOICE · SCORE · NOUL
Jev AI has three typed primitives. Choice selects from fixed options, Score rates an ordered scale, and Noul estimates a yes probability. Each constrains the answer differently, so you can match the type to the decision your application needs.
| Type | Returns | Rules | When to use it |
|---|---|---|---|
| Choice | One option, probabilities, and confidence | 2–255 predefined options | Routing, categorization, or selecting one next action |
| Score | A position on an ordered scale plus probabilities | 2–10 ordered levels; the score can be fractional | Severity, quality, sentiment, or priority |
| Noul | The probability that a statement is true | A value from 0 to 1 | Binary checks, gates, or verification |
JEV VS GENERAL-PURPOSE LLMs
Jev is optimized for focused decisions built with Choice questions, Score scales, and Noul probabilities. General-purpose LLMs are designed for open-ended generation, conversation, and multi-step reasoning.
| Dimension | Jev | General-purpose LLM |
|---|---|---|
| Output shape | Choice, Score, or Noul with probabilities | Generated text; structured output may be optional |
| Parsing | Answer shape is defined before the request | Often needs schema enforcement, validation, or text parsing |
| Latency | Designed for fast, focused decisions | Varies with model size and generated output length |
| Billing | Input tokens; decision output is currently free | Usually input tokens plus generated output tokens |
| Generates text | No | Yes |
| Best fit | Classification, routing, scoring, and verification | Chat, writing, code, explanations, and broad reasoning |
A constrained answer can still be wrong. Test on your own data and use confidence thresholds or human review for consequential decisions.
JEV API PRICING
Estimate monthly spend at current standard input rates for Jev, GPT-5.6 Terra, and Claude Sonnet 5.
Enter a token count or use K, M, and B shorthand.
Quick estimates
Estimated monthly input cost
At this volume
For 1M input tokens, Jev is 97.9% less and saves $1.96 versus GPT-5.6 Terra or Claude Sonnet 5.
Input-token prices only. Estimates use standard API rates and exclude LLM output tokens, caching, batch discounts, taxes, and provider fees.
Prices checked September 19, 2026.
COPY-PASTE JEV CODE EXAMPLES
These minimal examples follow the current provider documentation. Set the required API key in your environment before running them.
pip install typesafe-sdkfrom typesafe_sdk import Choice, TypeSafeClient
with TypeSafeClient() as client:
response = client.system_one(
state="I was charged twice. Please refund the duplicate.",
questions={
"route": Choice(
instructions="Which team should handle this?",
criteria={"billing": None, "technical": None},
)
},
)
print(response.choices["route"].choice)TypeSafe Python SDKnpm install @typesafe-ai/sdkimport { choice, TypeSafeClient } from "@typesafe-ai/sdk";
const client = new TypeSafeClient();
const response = await client.systemOne({
state: { document: "I was charged twice. Please refund it." },
questions: {
route: choice("Which team should handle this?", {
billing: null,
technical: null,
}),
},
});
console.log(response.answers.route.choice);TypeSafe JavaScript SDKpip install langchain-typesafefrom langchain_typesafe import Noul, TypeSafeClassifier
classifier = TypeSafeClassifier()
response = classifier.invoke(
state="The deploy failed twice. Customers are seeing 500s.",
questions={
"urgent": Noul(
instructions="Does this need attention right now?"
)
},
)
print(response.nouls["urgent"].noul)LangChain Jev guidenpm install aiimport { experimental_evaluate as evaluate } from "ai";
const result = await evaluate({
model: "typesafe-ai/jev",
state: "The support agent issued a full refund.",
questions: {
refunded: {
type: "boolean",
instructions: "Was a refund issued?",
},
},
});
console.log(result);Vercel Jev model pageJEV PLAYGROUND FAQ
Jev is a decision model from TypeSafe AI. It evaluates shared text or structured state and returns typed Choice, Score, or Noul answers with probabilities instead of free-form prose.
No. You can run limited shared requests here without an account or API key. To integrate Jev into your own application, you need your own API credentials.
Jev makes constrained decisions inside an answer shape you define. ChatGPT and Claude are general-purpose models designed to generate text, code, explanations, and conversational responses.
Choice selects one option from a predefined set. Score places the state on an ordered scale. Noul returns the probability that a yes-or-no statement is true.
The Jev Playground supports up to six questions about the same state. Sending them together avoids repeating the shared state in separate requests.
Jev cannot invent an option outside the answer shape you define, but it can still choose the wrong option or assign a misleading probability. Review confidence and test important workflows on your own data.
No. We do not store the state or questions you submit or include them in analytics.