Fast workflow branches
Ask the model to choose among approved paths and let deterministic application code perform the branch.
Decision model · Liquid
Route the next step with less output.
Liquid D1 is designed for the moment where your workflow already has context and needs a clear, machine-readable decision to continue.
Use the exact model ID in the request. Credits follow the same Jev rule as the other decision models.
A decision boundary
liquid/d1State in. Route out.
State
The account has an overdue invoice and a verified payment method.
Choice
Route to payment retry
Score
0.91 confidence
Noul
Do not contact if consent is missing
Keep the action in your application and the decision easy to audit.
Input
State + typed questions
Output
Choice · score · noul
Billing
Same as Jev
Endpoint
POST /v1/systemone
Why this model
A compact model for operational routing.
Use a small decision surface for queues, gates, and handoffs where the next step should be explicit.
Typed control signals
Keep output in choice, score, or noul form so downstream code does not need to interpret prose.
Inspectable state
Record the state, question, answer, and confidence together for debugging and review.
Where it fits
Use it at the handoff between judgment and action.
Liquid D1 works well when the set of valid next steps is known and the application owns execution.
PatternUse it forGood boundary
Queue routing
Choose a team, queue, or priority
Return an approved route
Policy gate
Check whether a rule is satisfied
Block or continue explicitly
Workflow handoff
Choose the next owner or tool
Keep permissions in your code
Call the model
Keep the request small and the result useful.
The same Jev API contract lets you compare Liquid D1 with the other decision models without rewriting your integration.
- Set model to the exact model ID shown below.
- Send the current state as a compact string or JSON-serializable value.
- Ask one or more typed questions and handle the returned answer in application code.
The model ID is liquid/d1.
Put the business context in state and the decision contract in questions.
Treat low confidence or a noul result as a first-class fallback, not as an exception.
Example request
curl -X POST https://thejevai.com/v1/systemone \
-H "Authorization: Bearer <API_KEY>" \
-H "Content-Type: application/json" \
-d '{
"model": "liquid/d1",
"state": "A customer request needs a safe next step.",
"questions": {
"route": {
"type": "choice",
"instructions": "Choose the correct next step.",
"criteria": {
"continue": "Continue the approved workflow",
"review": "Send the case for human review"
}
}
}
}'Keep the API key on your server. The Playground uses the same request shape through your signed-in session.
Build with it
Make routine decisions explicit.
Liquid D1 is a useful starting point for bounded, repeatable workflow decisions.
Support triage
Route a ticket to the right team while keeping the final reply under application control.
Payment safeguards
Check a compact set of conditions before retrying, refunding, or escalating a payment case.
Tool selection
Choose one approved tool path and let your permission layer decide whether to run it.
FAQ
A few practical integration questions.
How do I select Liquid D1?
Set model to liquid/d1 in the POST /v1/systemone request. You can also choose Liquid D1 in the Playground model selector.
Is Liquid D1 charged differently?
No. It follows the same Jev-based credit rule used by the supported decision models. Re-test latency and output quality with your own workload.
Should I let the model call tools?
No. Use the typed answer to select an allowed path, then let your application enforce authentication, permissions, rate limits, and side effects.
References
Make the decision contract easy to inspect.
OpenRouter model page
Check the current provider listing and availability for Liquid D1.
Jev API documentation
Follow the shared endpoint, authentication, and typed response format.
A representative test set
Keep examples for expected routes, ambiguous cases, and explicit escalation.
Start with a bounded question
Give every workflow branch a clear owner.
Try Liquid D1 with a small set of approved outcomes, then evaluate it against real routing cases.