Google Releases EmbeddingGemma 2: Unified Multimodal Vector Space

X postMiaAnnouncement

Google has launched EmbeddingGemma 2, an open-weights model mapping text, code, images, video, and audio into a single shared vector space. The 1.49 GB model is released under the Apache 2.0 licence.

Key points

  1. EmbeddingGemma 2 projects text, code, images, video, and audio into a single unified embedding space.
  2. Eliminates the traditional pipeline where media had to be transcribed or captioned before being processed by text embedders.
  3. Enables cross-modal querying, such as searching video clips with voice memos or scanning hours of audio with text prompts.
  4. The model weights measure 1.49 GB and are published as open weights on Hugging Face.
  5. Distributed under the Apache 2.0 licence, making it freely usable in commercial products.

Why it matters

Multimodal retrieval previously required complex glue code and separate speech-to-text or image-captioning stages. A unified vector space streamlines RAG architectures and cuts processing overhead. The lightweight footprint of 1.49 GB also makes local, on-device execution practical without transmitting raw media to third-party APIs.

Analysis

According to Mia, EmbeddingGemma 2 significantly simplifies multimodal RAG by doing away with transcription and captioning intermediaries. Having five modalities unified in a 1.49 GB model under an Apache 2.0 licence lowers the barrier for building cross-modal retrieval systems, local search applications, and native agent memory.

Original post

Mia

@MiaAI_lab · 6 October 2026

Google launched EmbeddingGemma 2 🔥 This is a bigger than you think. Most people build multimodal RAG as caption/transcribe, and then to text embedder. But EmbeddingGemma 2 puts text, code, images, video, and audio into ONE shared vector space, so you can query a voice memo t.co/O03JTlwtVK

858 likes36 replies684 bookmarks37k views

View on X
More posts in the thread (2)
  1. @niklaslenz_ai Yeah it's insane tbh
  2. btw the model weights only 1.49 GB !
Selected replies (5)
  • @niklaslenz_ai @MiaAI_lab For that size, pretty impressive.
  • @brnsantanna @MiaAI_lab A multi modal embeddings open source model is indeed valuable!
  • @mrfokki @MiaAI_lab one vector space for text, code, audio and video quietly kills a lot of ugly glue code
  • @MichaelHutu @MiaAI_lab the size of it 👀
  • @JulianPscheid @MiaAI_lab 100% Agree. I think this is the model we were waiting for to allow users to upload files and images into @HedyAI_ without the files leaving their local device.

Mentioned links

Links and tools from this entry

Summary generated by AI (Gemini 3.8 Flash, 7 October 2026). It may contain errors – the original source is authoritative.

Related entries

Similar in content, found by the AI search.

  • Article:Nikita Pavlichenko

    JetBrains Introduces Mellum2: A 12B Mixture-of-Experts Model for Code and Text

    JetBrains has released Mellum2, a 12B-parameter Mixture-of-Experts model trained from scratch on natural language and code. Activating only 2.5B parameters per token and published under the Apache 2.0 license, it is designed for low-latency, high-throughput developer and AI workloads.

    AI & LLMs· Announcement

  • X post:Xuan-Son Nguyen

    Local Desktop Search Tool Powered by EmbeddingGemma 2 and llama.cpp

    A desktop search utility shared on Reddit runs EmbeddingGemma 2 via llama.cpp to index and retrieve local files on macOS. Xuan-Son Nguyen highlighted the tool, calling it far superior to Apple's default Spotlight search.

    2.3kbookmarks260.3kviews

    AI & LLMs· Tool

  • Repository:google-ai-edge/LiteRT-LM

    LiteRT-LM: Google's Open-Source Inference Framework for On-Device LLMs

    LiteRT-LM is an open-source inference orchestration layer from Google designed to run large language models directly on edge devices. It leverages GPU and NPU acceleration and powers on-device generative AI features in Chrome, Pixel Watch, and Chromebook Plus.

    6.4kstarsC++

    AI & LLMs· Tool

Back to the knowledge base

Let's talk about your project

Get in touch

Locations

  • Mattersburg
    Johann Nepomuk Bergerstraße 7/2/14
    7210 Mattersburg, Austria
  • Vienna
    Ungargasse 64-66/3/404
    1030 Wien, Austria

Parts of this content were created with the assistance of AI.