Text & documents
Numbers & maths
Data & formats
Security
Development & DevOps
Artificial Intelligence
Finance
Health & Wellness
Productivity
Games & Entertainment
Multimedia & design
Business
How to use
How it works

Computes the semantic similarity between texts using the paraphrase-multilingual-MiniLM-L12-v2 model, which turns each text into a 384-dimension vector (embedding). Cosine similarity measures how alike they are: 1.0 = identical meaning, 0.0 = unrelated. The model runs 100% in your browser (WebAssembly); no text leaves your device.

Compare 2 texts

A direct similarity score between two texts. Useful for measuring how semantically alike two sentences or paragraphs are.

EmbeddingsSimilarity, 2D map and RAG
Semantic similarityCosine similarity with local AI · How RAG works