Prepaid balance, prices shown in toman, and how the cost of each request is calculated.
DaLAi has no monthly subscription. You have a wallet that is topped up in advance, and each request deducts from it according to the tokens it actually used. If you send nothing, nothing is deducted.
Cost is calculated per token, not per number of requests. Each request has two numbers:
One important point that explains most users' costs: in a multi-turn conversation, the entire history is sent again each time. That is why the tenth message costs more than the first, even if it is shorter.
Prices are shown per one million tokens. As a rule of thumb, a typical message is about a thousand tokens, that is, about one-thousandth of the figure you see in the list.
Internal accounting is in dollars, because the models themselves are priced in dollars. What you see is in toman, and the exchange rate is updated daily.
In practice: a model's toman price may change slightly from one day to the next without its dollar price having changed. The balance you bought does not shrink; only its display is calculated at the day's rate.
The live price of every model is in Model Square. The figure you see there is exactly what you pay.
We deliberately do not include a price table in this documentation: the rate changes daily, and a fixed table in the text would be wrong within a day.
In the Console:
If a figure looks wrong to you, that request's row in Usage Logs shows the full breakdown.
To add to your balance, contact us. An automatic payment gateway is not active yet and is being worked on; until then, top-ups are made manually, after coordinating with us.
If your balance runs out, requests return an error, and its message may be in English or Chinese. It means exactly that: your balance is insufficient. See Errors and limits for details.