Tovanix
Tovanix

AI Workload Routing Lab

Declare your workload constraints and get a reviewable, reproducible candidate routing policy. This page only produces candidates: no policy takes effect and no gateway configuration is changed.

  • Catalog source: a public read-only projection with catalog version, evidence timestamp and digest; a stale catalog is flagged as such
  • Workload constraints: protocol, minimum context window, required capabilities, data residency and fallback tolerance
  • Candidate result: primary and fallbacks, the constraints that applied, rejected models with reasons, and unresolved items listed one by one
  • Cost: the page never estimates an amount; pricing must come from a Quote Passport, which is not yet available
  • Export: inputs, catalog and rules versions, evidence timestamp and the not-activated status, ready to paste into a review record

Frequently Asked Questions

Does a candidate policy from this page take effect automatically?

No. The page only produces candidate suggestions; it activates nothing and changes no gateway configuration. To actually use one, configure it in the console by hand according to your review.

Why is there no cost shown?

Pricing must come from a Quote Passport, and that source is not yet available. Entering a budget cap does not compute an amount either — it only makes the gap show up explicitly as an unresolved item instead of a made-up estimate.

Related pages: Tovanix API Documentation · AI API Gateway · Direct LLM Access · Using the Astrenix AI gateway

Loading…