Rates in the open

One key.
Every model, priced.

These are the rates a call is billed at, taken from the same catalogue the router reads. Open any row for the per-route detail; nothing is marked up between the provider and your balance.

Model calls

Per token

Charged at the catalogue rate for the model that actually served the request. Nothing is added on top on the way out.

  • Input and output metered separately
  • Rate shown before the call, not after
  • Every request itemised in the usage log

Agent runtime

From $5 / month

A managed runtime that stays online without a machine of your own. Attach a key and the model usage bills separately.

  • Flat monthly line per running agent
  • Parked agents drop to nothing
  • Automatic rebuild on failure

Linux workspace

From $15 / month

A dedicated VM with the desktop and one bundled runtime on it. No second hosting charge for the agent living there.

  • 6 vCPU · 12 GB RAM · 200 GB NVMe
  • Nine regions to choose from
  • Model tokens still metered per call

Catalogue

Model rates

Filter by type, search by model id or vendor. Prices are per million tokens.

Sample dataa representative slice of a catalogue, not a live price feed.

ModelTypeInput / 1MOutput / 1MRoutes

atlas-6-wide

Northwind

Multimodal$0.55$2.769

lumen-5-pro

Calder

Multimodal$1.32$6.627

swift-3.8-flash

Meridian

Multimodal$0.08$0.419

atlas-5.6-solo

Northwind

Multimodal$0.28$1.6510

delta-v4-flash

Kestrel

Chat$0.33$0.993

flare-2.5-image

Northwind

Image$1.65$9.905

quill-4.6

Sable

Multimodal$0.22$0.663

kite-k3

Harbour

Chat$2.25$11.254
16 of 16 sample rows1 / 2