A small AI operation, such as categorizing the sentiment of a piece of feedback, using a standard AI model, such as GPT-4o or Claude 3.5 Sonnet, typically equates to two credits.
A small AI operation like that mentioned above, using a “lite” AI model, such as GPT-4o mini or Claude 3.5 Haiku, typically equates to one credit.
The size of the operation is determined by the number of characters in the request (the “prompt”), combined with the number of characters in the AI response.
Larger inputs (like documents, call transcripts, or web searches) will consume proportionally more credits than smaller inputs (like a project description), and larger outputs (like having AI produce long documents or blog posts) will consume proportionally more credits than smaller outputs (like a category or label).
When creating an AI field, you’re able to see a preview of the exact number of credits your prompt and input data will consume. You can estimate credit usage for a use case by multiplying this by the number of records you expect to generate/refresh.
Airtable AI’s generate formula, comment summary, and suggested matches in linked features all use one credit every time they’re used for customers accessing free credits. For those with the paid AI add-on, these features are available on an unlimited basis and do not consume your credit balance.
We encourage you to try out our free credits to get a sense of how many credits will be needed to accomplish your workflows.