# The Decoder — Google claims EmbeddingGemma 2 outperforms rival embedding models twice its size

- Company: The Decoder (the-decoder.com)
- Announced: 2026-10-06T19:47:27+00:00
- Category: open-weight-release
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://the-decoder.com/google-claims-embeddinggemma-2-outperforms-rival-embedding-models-twice-its-size/
- Record: https://forck.live/items/16260-google-claims-embeddinggemma-2-outperforms-rival-embedding-models-twice-its-size
- Subject: Reporting
- Open weights: yes
- Models affected: EmbeddingGemma 2, Gemma 4

Google released EmbeddingGemma 2, an open multimodal embedding model with 740 million parameters that converts text, images, video, audio, and code into numerical vectors. The company claims it outperforms competing models up to twice its size on multimodal embedding benchmarks while running locally without an API key, requiring about 191 MB of RAM and taking 20–70 milliseconds per query via WebGPU in the browser. A 270-million-parameter text-only version is also available.

## Evidence

Verbatim from https://the-decoder.com/google-claims-embeddinggemma-2-outperforms-rival-embedding-models-twice-its-size/:

> At 740 million parameters, Google says it's the most compact model of its kind and outperforms competing models up to twice its size on multimodal embedding benchmarks.

---

Record: https://forck.live/items/16260-google-claims-embeddinggemma-2-outperforms-rival-embedding-models-twice-its-size
Catalogue: https://forck.live/llms.txt
Current issue: https://forck.live/feed.md