text Faster on WebGPU

Text Embeddings

Compute semantic embeddings and pairwise similarity for text files — runs on your device.

First run downloads ~22 MB. The model is cached after the first use, then runs offline. Manage downloads on the settings page.
Loading…

About Text Embeddings

Text Embeddings turns text into semantic vectors and computes pairwise similarity between passages, all on your device. Use it to measure how close two pieces of text are in meaning — for clustering, dedup, or building the scoring layer of a small semantic-search setup. The text and the resulting vectors stay in your browser and are returned as JSON.

Category
text
Input
Accepts: text/plain.
Output
Outputs: application/json.
Cost
Free, runs in your browser
Memory
low
Install group
nlp-standard
Privacy: Text Embeddings runs entirely on your device. Files you provide never leave your browser — no uploads, no server, no tracking. The page works offline once loaded.

Common uses

  • Measure how semantically close two paragraphs are even when they share no exact words
  • Cluster a set of support messages or survey answers by meaning
  • Build the similarity-scoring step of a lightweight semantic search over your own notes
  • Detect duplicate or near-duplicate FAQ entries that are worded differently
  • Rank candidate snippets by how well they match a reference passage
  • Find the most related document in a small collection given a query sentence

Frequently asked questions

What exactly does it output?

JSON containing the embeddings and pairwise similarity for the text files you provide, so you can use the numbers programmatically.

Does my text leave the browser?

No. The embedding model runs locally on your device, so neither the text nor the vectors are uploaded.

What is a semantic embedding?

A numeric vector that captures the meaning of text so that passages with similar meaning land close together, even with different wording.

What input does it take?

Plain text. Provide the passages you want embedded and compared.

How is this different from Image Similarity?

Same idea, different medium: this compares the meaning of text, while Image Similarity compares the visual content of images.

Can I use these vectors elsewhere?

Yes. The JSON output is portable, so you can store the embeddings or feed the similarity scores into your own pipeline.

Keywords

  • embed
  • embedding
  • semantic
  • similarity
  • sentence
  • vector
  • nlp
  • search

Try next