Skip to main content
Use aiStats.textEmbeddingModel(...) (or embeddingModel) with embed and embedMany. 
import { aiStats } from "@ai-stats/ai-sdk-provider";
import { embed, embedMany } from "ai";

const single = await embed({
  model: aiStats.textEmbeddingModel("google/gemini-embedding-001"),
  value: "Embeddings power semantic search.",
});

console.log(single.embedding.length);

const batch = await embedMany({
  model: aiStats.textEmbeddingModel("google/gemini-embedding-001"),
  values: ["pricing", "latency", "quality"],
});

console.log(batch.embeddings.length);

​
Notes

  • Keep query and corpus embeddings on the same model version.
  • Normalize vectors consistently before similarity scoring.
  • Cache stable embeddings for repeated content.
Last modified on March 16, 2026