You pay for model requests
Requests to Liho LLM models are paid from a ruble balance at catalog prices, whether they come from the API, Cowork or Liho IDE. You top up the balance in advance and are charged for the actual usage of each request.
Output token prices in the catalog range from ₽0.4136 to ₽165.44 per 1M.
What you pay for
| What | How it is paid |
|---|---|
| Model requests through Liho LLM: API, Cowork, Liho Server, the Liho IDE agent | From the Liho LLM account balance at catalog prices, with input and output tokens priced separately |
| Liho Cowork and Liho IDE installers | Free to download. Costs arise when the apps make model requests |
| Your own Liho Server | You rent the server from your own hosting provider. Model requests come from the same Liho LLM balance |
| Liho Cloud | The workspace is created in Cowork settings. Model requests are paid from the Liho LLM balance |
| Your own provider keys and CLI subscriptions in Liho IDE | Paid to those providers, not through Liho |
How the cost of a request is calculated
Catalog prices are per 1M tokens and already include the service fee.
cost = input tokens × input price + output tokens × output price
- If a model has a cached input price, cached tokens are billed at that price.
- Reasoning tokens are billed as output tokens.
- The catalog price at the time of the request applies.
- The cost comes back in the API response in the usage.cost field and is shown in the request history.
Example: GPT-6 Sol
10,000 input and 2,000 output tokens: ₽0.2482 + ₽0.0993 = ₽0.3474.
Prices: ₽24.82 / ₽49.63 per 1M input / output tokens.
The reserve before a request
Before a request, the maximum possible cost is reserved from your balance, based on the request length and max_tokens. After the response, the actual amount is charged and the rest of the reserve is returned right away. The dashboard shows three values: balance, the reserve for requests in progress, and the available amount.
If the outcome of a request at the provider is unknown, the reserve is held until reconciliation and is not returned automatically after a timeout.
Spending control
- A separate API key for each app, server or employee.
- A spending limit per key: lifetime, or reset every day, week or month.
- A requests-per-minute limit and a key expiration date.
- Request history with filters by key and model, and CSV export.
Balance
The service is prepaid: you top up the balance in rubles in the dashboard before use. The unused balance is not refunded except as required by law; see the Terms of Service for details.
Liho LLM Terms of ServicePricing questions
No. You pay for actual model requests from your balance.
Liho LLM prices are in rubles and include the service fee. Current values are in the catalog.
A request that did not complete is not charged: the reserve is released. If the outcome at the provider is unknown, the reserve waits for reconciliation, and the request is marked as pending in the history.
Not at the moment: you can pay only by topping up your balance in the dashboard.
Compare models by price
The catalog lists prices for all models per 1M input and output tokens.