EmbeddingGemma 2

Try google/embeddinggemma-2, Google DeepMind's open embedding model, running entirely in your browser. Nothing you type or upload leaves your device. Each tab can shorten embeddings from 768 to 512, 256 or 128 dimensions (Matryoshka) so you can compare results.

Starting…

Task

Query

Documents (one per line)

Embedding dimension

768 512 256 128

Ranked documents

Runs onnx-community/embeddinggemma-2-ONNX (4-bit) with Transformers.js. The first visit downloads about 210 MB; later visits load from your browser's cache. Text uses the model's task prefixes (e.g. task: search result | query: …); images use none.