How much will a bot that knows my documents cost?
Want a chatbot that talks to your files (FAQs, docs, product catalog, manuals)? First you need to convert each document into a format the bot can understand. The operation is called embedding. It costs pennies, but with thousands of files it pays to know exactly how much.
Enter: how many documents you have, average length, how often you want to refresh. The calculator shows today's cost and yearly cost for 5 popular services (OpenAI, Voyage, Google).
Plus ready scenarios: 10k FAQs, 5k pages of docs, 100k product catalog, 50k support tickets, 100 books. Click, see what you'd pay for your case.
How to use it
- Click a scenario close to yours (FAQ, docs, store, support, books), or type your own numbers.
- Re-indexing per year slider = how often you want to redo all documents. 0 = never (data doesn't change). 1 = once a year. 12 = monthly.
- The table shows one-time cost and yearly cost for each service.
- Cheapest is highlighted green. Below each is a short note on when to pick it.
- USD/EUR toggle: show in EUR (set the rate manually).
When this is useful
Six typical situations where the calculator gives you a concrete answer instead of a guess:
- Quoting a client project. The client has 5,000 PDFs of 1,000 words each. They ask: *"what will it cost to make the bot 'read' them?"*. The calculator says: ten cents. Concrete number instead of guessing.
- Picking a provider. All services do the same thing, but differ in price and quality. The cheap one is enough for 80% of use cases, 6× cheaper than the premium one. The premium one only pays off for long, technical documents.
- Updating the catalog monthly. A store with 100k products, prices change every month. The calculator says: a few dollars per year. Trivial cost, nothing to debate.
- Migrating to a better service. You're using an old, expensive one. The newer one is 5× cheaper and better. The calculator shows the cost of a one-time re-conversion of all documents. Decision makes itself.
- Setting a budget. You want to know your company's total spend on *"teaching"* the bot. List everything the bot should know, the calculator sums it up.
- Comparing "small" vs "large". Is it worth paying 6× more for the bigger version? The calculator shows the difference in dollars for your data, easier to make an informed choice.
Questions and answers
An embedding is converting text into a format the bot can understand. Think of a cookbook. The bot doesn't read it like a human. You first convert each recipe into a special representation (that's the embedding). When you ask *"what to do with leftover chicken"*, the bot finds the representation closest to your question. That's the whole trick.