The Embedding Cost Calculator compares what embedding APIs cost across providers. You enter monthly input and output token volumes and the table returns input cost, output cost and a monthly total for each provider and model. It suits teams planning semantic search, recommendations or RAG indexes, where embedding every document can dominate API spend. Results are estimates based on published per-token pricing and update as you edit the inputs. No sign-up is required.
Embedding calls return vectors rather than text, but the table uses the same input and output structure as other providers so you can compare consistently. For embedding-only workloads, leave output at zero and read the input column. If you also generate text with the same provider, add those output tokens.
A rough rule is one token per four characters of English text, so a large document collection quickly reaches millions of tokens. Embeddings are usually computed once, then re-run when content changes. Estimate initial indexing and monthly refresh separately, and include query embeddings, which are small but frequent.
Match the model to your language coverage, vector dimensions and retrieval quality needs. Multilingual corpora need a multilingual model, and larger vectors cost more to store. Test candidates on a labelled retrieval set. Plugsky offers embedding models through an OpenAI-compatible API, including on the free plan with 2 free models.
Canonical pricing and plans: plugsky.com/#sec-pricing · Terms · SLA · Docs