Skip to content
Coming soonExplore the public preview. Connected model, compute and payment services are in development.Service status
SAVRN
Search Contact SAVRN
SAVRN Cloud

Reference

Schemas and service limits

A field's unit, revision and meaning are as important as its displayed value. This SAVRN Cloud guide uses labeled sample records to explain the difference between an illustrative schema, a configured allowance and a supported service limit, including how unknown measurements should remain explicit.

Download Markdown

SAVRN Cloud · Coming soon · Public preview · Browser-local simulation · No live compute or payments

Read local records carefully

Inspect a download from Usage, Deployments or another screen that provides an export. Keep its simulation metadata and distinguish display values from real measurements. Browser-local record formats may change while this demonstration develops; they are not a published production schema.

The following shows a documentation principle, not a supported import format:

{
  "simulation": true,
  "schema_status": "illustrative",
  "model_name": "Research 8B",
  "production_measurements": null,
  "qualified_production_offers": [],
  "capacity_verified": false
}

An empty offer list means none are established here. A null measurement means unknown, not zero. A numeric fixture is not a benchmark.

Define every limit

A production parameter needs its type, unit, allowed range, default, required/optional status and error behavior. Distinguish hard implementation limits from configured project allowances and currently available capacity. Record whether a duration concerns queueing, execution or total lifetime.

Before connected services launch

Generate reference tables from the supported schema and approved offer revisions. Version changes and document compatibility. Test boundary values, missing fields and unsupported combinations. Keep secrets, private addresses and tenant-only attributes out of public projections. Add freshness and expiry to capacity observations so stale data cannot silently continue to advertise availability.

Build toward the work that matters.

Tell SAVRN what your institution needs to run, who reviews the results and where its data must stay. That workload defines the next service to qualify.

Discuss an AI project