API pricing and charges in Tokeniser

Tokens, model multipliers, minimum charge, reservations and tariff versions.

Top up your balance

Choose a package or a whole-rouble amount starting at 100 RUB. Before payment, check the final discounted amount, credited tokens and price per million tokens. Discounts and refunds may include kopecks. The model table shows ordinary and cache multipliers.

Funding and usage

A top-up credits tokens under the selected offer. Requests spend tokens from your balance at the model multiplier. Check the payable amount and credited tokens before payment.

How ×N works

A uniform tariff applies one final model multiplier to requests, responses and reasoning. At ×2, 1,000 ordinary tokens spend 2,000 tokens from your balance; 1,000 provider-confirmed cached input tokens spend 800: cache costs 40% of the ordinary rate. Cached input is excluded from ordinary input; reasoning is not added to output again. The total is rounded up once, then the displayed request minimum applies. Legacy split tariffs are not presented as a single ×N.

Reservations and versions

Accepted requests retain their tariff version. An upper usage estimate is reserved, then the actual amount is charged and the remainder released. See the catalog for final rates and minimums.

Images and audio

Separately accepted modes can charge per image or audio second with specific allowed settings. Ordinary text-token rates do not describe those charges.