Model catalog

LunaRoute serves a curated roster of frontier models — models we run, certify, and support ourselves. Every model is available on every plan, behind the same endpoint; the plan changes your lanes, not which models you can call.

Routing through your own provider keys is supported as a secondary path, but the curated roster is the managed default.

Query the catalog

The catalog is live data, not a static list in these docs. Always read it from the API:

curl https://gw.lunaroute.com/v1/models \
  -H "LUNAROUTE-API-KEY: $LUNAROUTE_API_KEY"

The response is the platform’s global catalog — the models with a configured provider credential, the same entries for every authenticated caller. It shows what the platform serves, not a per-organization entitlement. The CLI shows the same catalog:

lunaroute models     # available models
lunaroute pricing    # per-model credit pricing

See the Models API for the response shape.

Variants

A model entry may expose quantization variants — the same base model at different precisions (for example an nvfp4 variant). Variants differ in throughput and cost, not in API shape; select one by its model id exactly as listed.

Some models also carry per-client compatibility metadata, so an agent client gets the right defaults for its dialect.

Pricing

Pricing lives in the catalog and the usage ledger, not in these pages — prices change and are pinned per model at request time. Use lunaroute pricing or the dashboard ledger for current rates.