The AI tools in Stella interact with commercial LLM providers who charge for activity based on the number of tokens sent, processed, and received back. The cost of a query, thus, is roughly proportional to the amount of information sent and the volume of content sent back. The sd-ai Gateway is designed to minimize token use while not degrading quality. This is one of the purposes of the AI Tools and Actions which allow the LLMs to work on parts of models and results.
Still, even the most efficient interactions with LLMs incur costs, which vary by LLM provider. Stella provides a convenient way to purchase credit so that you do not have to deal directly with API Keys and billing from LLM providers. In a nutshell, the default sd-ai Gateway keeps track of usage and deducts the costs of queries from the AI Credit Balance associated with your registration code. When your AI Credit Balance is depleted, you will no longer be able to use the AI tools.
You can check your AI balance anytime from the AI Assistant Credit dialog from the AI Assistant by clicking on the $ button. As you make queries, the usage will be recorded and the account credit will be decreased. You can determine the cost of a query by looking at the balance before and after making a query. The balance should update within a minute or two of the query completion.
If you want to set up your own API keys with different providers the best way to do this is to configure an sd-ai Gateway to run locally. Though you will still see the $ button to check balances, there will be no effect on these from using the AI tools, these will be billed directly against your API keys. The gateway also have a cost function that will allow you to monitor usage in real time if that is desirable.