Embedding Cost Calculator

Cost equals tokens divided by one million, multiplied by the price per 1M tokens, multiplied by the number of passes. Embedding APIs bill only on input tokens — there are no output tokens — so a single price per 1M tokens drives the whole estimate.

If you know the token count, enter it directly. Otherwise paste a representative sample of your text and the tool estimates tokens at roughly four characters per token. That heuristic is approximate; the real count depends on the model's tokenizer, so treat it as a ballpark.

You often embed the same corpus more than once — re-indexing after content changes, A/B testing chunk sizes, or rebuilding a vector store. The passes field multiplies your per-pass token volume so the total reflects every time the data is sent to the model.

The presets are illustrative example prices, not live quotes. Embedding pricing changes and varies by provider and model, so the price field is editable — set it to your provider's current per-1M-token rate before relying on the estimate.