Skip to content
🏡 Start Here


Embedbase is a single API to access both LLMs and a VectorDB*

Key features

  • Generate: use .generateText() to use 5+ LLMs
  • Semantic Search: use .add() to create a list of semantically searchable information and .search() to run semantic queries


Here's a small example to do a simple Q&A search app:

import { createClient } from 'embedbase-js'
// initialize client
const embedbase = createClient(
  '<grab me here>'
const question =
  'im looking for a nice pant that is comfortable and i can both use for work and for climbing'
// search for information in a pre-defined dataset and returns the most relevant data
const searchResults = await embedbase.dataset('product-ads').search(question)
// transform the results into a string so they can be easily used inside a prompt
const stringifiedSearchResults = searchResults
  .map(result =>
const answer = await embedbase
  .useModel('openai/gpt-3.5-turbo-16k') // or google/bison
  .generateText(`${stringifiedSearchResults} ${question}`)
console.log(answer) // 'I suggest considering harem pants for your needs. Harem pants are known for their ...'

Checkout the .add() documentation to see how to populate the dataset.


npm i embedbase-js

Learn more

SDKs documentation (opens in a new tab)The Embedbase JS and Python SDKs
Examples (opens in a new tab)Try some examples