Models & pricing

This page reads GET /v1/models (the same catalog billing uses) and lists the current retail price per million tokens (USD) for each model in the catalog. Prices are frozen when the request’s hold is created, so later price changes never affect a request already sent; the receipt for that request is authoritative.

The pricing policy is shown per model: under openrouter_reference_markup_v1, the retail price is the upstream reference price frozen at review time plus that model’s markup (markup_bps, for example 1200 = +12%); legacy policies already include their fees in the retail price. What a user actually pays can also include consented app service fees (not collected yet) and the payment processing fee charged at top-up.

Available means current pricing permits new requests, not a guarantee of upstream success. Status refreshes every 30 seconds; expired models remain listed for reference.