Knowledge base storage is metered in chunks—the AI-searchable content extracted from your documents, each chunk a small piece of extracted text (about 1.3 KB). The “≈ documents” figures assume typical business files, where 1 GB of documents yields about 120,000 chunks. Text-dense files consume quota faster; low-text files (scans, images) consume less, so many customers fit more than the headline figure. You only pay for the text that is extracted; not the size of the file. Your original files are kept and never count against quota.
An upload that would take you past your included storage is held until you approve the additional storage (one click); approved storage overage is billed on the highest level reached during the cycle—deleting documents frees quota immediately but does not reduce that cycle’s overage.