Live per-token rates for every model on the platform. Updated automatically from upstream provider catalogs.
Per-request transaction model.
Each API request is a discrete transaction priced at the rate captured by our pricing service at the moment of metering.
Rates listed here are informational and refresh from upstream catalogs periodically (up to 24 hours of lag).
Customers are charged at the rate in effect when the request is metered, not the rate displayed at submission time.
See Terms §3 for the full description, including the manifest-error and rate-source clauses.
Model
Provider
Tier
Residency
Input CA$/M
Output CA$/M
Cache read CA$/M
Cache write CA$/M
Loading…
Per million tokens. Cache rates apply only when the model + your client both support context caching;
otherwise input rate applies. Customers are billed in Canadian Dollars (CAD).
Upstream provider rates are published in USD; the CAD column converts at the rate shown above for display purposes.
Actual customer charges convert at metering time using Stripe's published FX rate (per Terms §3).
Sources of truth. Bedrock rates come from the AWS Pricing API + AmazonBedrockFoundationModels offer file.
Azure rates come from Azure Retail Prices. Direct-API providers (Anthropic, OpenAI, Cohere, Mistral, etc.) come from OpenRouter
mirroring the publisher's own page. Each rate carries its source provenance internally; admins can see the upstream URL + cache age
via the admin UI. See our Acceptable Use Policy and Privacy Policy.