Estimate what it costs to embed your corpus across OpenAI, Cohere, and Voyage. Adjust token volume and re-embed frequency — the estimate updates live.
Embedding providers bill per 1,000 input tokens with no separate output rate, since the API returns a fixed-size vector rather than generated text. This calculator multiplies your tokens-per-run by the provider's per-1K rate to get a cost per run, then multiplies by how many times per month you re-embed.
It doesn't account for batch API discounts, provider free tiers, or volume-tier pricing that kicks in above certain monthly thresholds. Treat this as a directional ceiling for planning, not an exact invoice.
Cost = (tokens embedded ÷ 1,000 × per-1K rate) × times re-embedded per month. Most providers bill per 1,000 input tokens with no output cost.
Teams re-embed when source documents change, when switching to a higher-quality model, or when rebuilding an index after a schema change.
Yes. Small models cost roughly 6x less per token than large models, and for many retrieval tasks the quality gap is small enough to be worth testing.