# Semantic Search Over Your Own Text, In the Browser

> Embed a list of sentences with a real model and rank them by meaning — or find the near-duplicates in a list.

This tool embeds each line of your text with a sentence-transformer model running in your browser, then ranks lines against a query by cosine similarity, finds near-duplicate pairs above a threshold, or groups lines that mean the same thing. Because it compares meaning rather than words, it matches "when are you open?" to "what are your business hours?" — and none of the text is uploaded.

**URL:** https://convertto.tech/t/semantic-search-tool
**Category:** On-Device AI (https://convertto.tech/c/local-ai-tools)
**Privacy:** Runs entirely in the browser; no upload
**Cost:** Free, no sign-up
**Last updated:** 2026-08-01

## Key facts

- **Uploads:** None — the model is downloaded to you, not the reverse
- **Runs on:** WebGPU where available, WebAssembly otherwise
- **Dimensions:** 384 on MiniLM, 768 on BGE-base, 1024 on mxbai-embed-large
- **Languages:** English, or 50+ with the multilingual model
- **Practical limit:** A few hundred lines per run
- **Privacy:** Runs entirely in your browser — nothing is uploaded
- **Cost:** Free, unlimited, no sign-up

## How to use

1. Enter or paste your text.
2. Choose the do what.
3. Enter or paste your query.
4. Choose the model.
5. Set the similarity threshold.
6. Set the results.
7. Press Run, then download the result when it is ready.

## FAQ

### How is this different from Ctrl+F?

Ctrl+F matches characters. This matches meaning: "my card was declined" and "payment failed at checkout" have no words in common and are recognised as the same complaint. That is the whole difference between keyword search and semantic search.

### Does it work across languages?

With the multilingual model, yes — it maps 50+ languages into one shared space, so a question in Spanish will match an answer written in English. The English-only models will not do that: they score a cross-language pair little better than an unrelated sentence.

### Is the largest model always the best choice?

For retrieval quality it is the strongest here, but it is a 321 MB download and slower per sentence. MiniLM is enough for deduplicating a list or searching a few dozen items; reach for the larger models when near-misses matter and the small one is confusing things that merely share vocabulary.

### What threshold should I use for duplicates?

Start at 0.75 and adjust while looking at the results. There is no universal figure — different models compress the similarity range differently, so a threshold copied from a tutorial written for another model will be wrong.

## Sources

- [Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks](https://arxiv.org/abs/1908.10084) — arXiv

## Related tools

- [Embedding Similarity Calculator](https://convertto.tech/t/embedding-similarity-calculator): Paste embedding vectors and get every pairwise cosine, dot product, Euclidean and Manhattan distance, ranked.
- [Chat with PDF](https://convertto.tech/t/chat-with-pdf): Ask questions about any PDF and get answers with the page each one came from. The model runs in your browser — the file is never uploaded.
- [Text Chunker for RAG](https://convertto.tech/t/text-chunker-for-rag): Chunk by tokens, sentences, paragraphs, a separator ladder or markdown headings — with overlap, token counts and embedding cost.
- [Chat with Any Document](https://convertto.tech/t/chat-with-document): Ask questions about a PDF, Word file, spreadsheet, CSV, Markdown or text file and get answers with the passage each came from.
- [Text Similarity Checker](https://convertto.tech/t/text-similarity-checker): Compare two texts with Levenshtein distance, Jaccard, cosine and Dice coefficients.
- [AI Alt Text Generator](https://convertto.tech/t/ai-alt-text-generator): Describe an image with a captioning model running in your browser — a first draft for the alt attribute you then edit.
- [AI Audio Transcription](https://convertto.tech/t/ai-audio-transcription): Run Whisper locally to turn speech into text, SRT, VTT or a timestamped transcript. Works offline after the first run.
- [AI Background Remover](https://convertto.tech/t/ai-background-remover): Cut the subject out of a photo with a real segmentation model that runs in your browser. No upload, no sign-up, no watermark.