EmbeddingGemma 2: Google's multimodal embedding model runs locally
Google DeepMind has released EmbeddingGemma 2, an open embedding model built on Gemma 4. It maps text, code, images, video and audio into a single shared 768-dimensional vector space and is designed for laptops and smartphones.
What it's about
Google DeepMind has released EmbeddingGemma 2, an open multimodal embedding model based on the Gemma 4 architecture. An embedding model turns inputs into lists of numbers (vectors) so that similar content ends up close together. That's the foundation for search, classification and clustering. EmbeddingGemma 2 is available in the Ollama library (ollama run embeddinggemma-2) in sizes 270m, 440m, 570m and 740m.
Architecture and capabilities
The model maps text (including code), images, video and audio, as well as combinations of them, into a single shared 768-dimensional vector space. That means you can, for example, use a text query to search for images or audio clips. In total it has 740M parameters (the settings learned during training): a 270M text model (130M transformer plus 140M embedder) and modular encoders for vision (170M) and audio (300M). These encoders can be loaded selectively, so you only use the modalities you need. That probably explains the tiered sizes in Ollama, but the source doesn't explain how they map.
Other properties according to Google:
- Understands more than 100 languages and is about 14% better than its predecessor on code tasks.
- 8K-token context window, enough for minutes of audio or video.
- Task-specific embeddings via short text prefixes (instructions) for search, classification, clustering or semantic similarity.
Matryoshka truncation
EmbeddingGemma 2 supports Matryoshka Representation Learning (MRL): like Russian nesting dolls, truncated vectors stay usable. You can cut them to 128, 256 or 512 dimensions and re-normalize. That saves up to 6x storage for vector databases. According to the source, quality loss is minimal down to 256d, and 128d works best for text-only tasks.
Benchmarks
The model was tested on text, code, vision, visual document, video and audio benchmarks, in each case using the full-precision checkpoint. The actual numbers from the tables aren't in the source we have, so we can't quote any figures. All claims come from Google.
What you need to try it
Google is targeting consumer hardware such as smartphones and laptops. The smallest variant (270m) runs on ordinary laptops without a dedicated GPU, and even the 740m model is small (roughly 1.3 to 1.5 GB according to the Ollama tags). Ollama runs on Mac, Linux and Windows, including CPU-only. The source gives no exact figures.
Our take
A single model for text, images, video and audio greatly simplifies multimodal search and RAG (answers generated with the help of retrieved documents), especially when run locally and with privacy in mind. The small size and flexible vector length cut costs. What remains unclear is how it stacks up against competing models, since the benchmark numbers are missing here. The license and details on using it in Ollama (how the sizes map) aren't given in the source either.
Sources
- Ollama-Bibliothek ollama.com
Try it yourself: EmbeddingGemma 2
Explained step by step, no prior knowledge needed – on your own computer or free in your browser.
What can I do with it?
EmbeddingGemma 2 turns text, code, images, video and audio into vectors. You can use it to build local semantic search, RAG, classification or clustering, even across different media types.
Who is it worth it for?
A good fit for developers who want to build local multimodal search or RAG. If you just want a chatbot, you don't need it, since it doesn't generate text.
What your computer needs
Which variant is right for me?
Many models come in several sizes or "quantizations." These are compressed versions: smaller and faster, but slightly less accurate. Rule of thumb: pick the largest variant whose memory requirement fits in your VRAM (video memory). The scripts below choose this for you automatically.
| Variant | Download | Needs approx. | Compatible graphics cards |
|---|---|---|---|
270m-bf16-text |
1.2 GB | 2.5 GB | almost any graphics card with 4 GB or more (e.g. GTX 1650, RTX 3050) |
270m Default |
1.3 GB | 2.5 GB | almost any graphics card with 4 GB or more (e.g. GTX 1650, RTX 3050) |
740m-q8_0 |
1.3 GB | 2.5 GB | almost any graphics card with 4 GB or more (e.g. GTX 1650, RTX 3050) |
740m-bf16 |
1.5 GB | 3 GB | almost any graphics card with 4 GB or more (e.g. GTX 1650, RTX 3050) |
Available sizes are 270m, 440m, 570m and 740m. The 270m model is the text base; based on the architecture, larger variants include the additional vision and audio encoders. The source doesn't state the exact mapping.
- Designed for laptops and mobile devices; no large VRAM needed.
- Vision (170M) and audio (300M) encoders can be loaded selectively to save memory.
- Shorter vectors (e.g. 256d) cut storage needs in the vector database by up to 6x.
- Context: 8K tokens.
How do I find out what my computer has?
- Windows: Press Ctrl + Shift + Esc → Task Manager → Performance tab → click GPU on the left. At the bottom you'll see Dedicated GPU memory (that's your VRAM). You'll find system memory under Memory.
- Mac: Apple menu at the top left → About This Mac. You'll see your chip (e.g. M2) and memory (e.g. 16 GB). On Apple chips, graphics and processor share this memory.
- Linux: Type
nvidia-smiin the terminal (shows your graphics card and its memory) andfree -hfor system memory.
Guide
Google Colab (in your browser, no installation)
Google Colab is a free service from Google. It gives you a computer with a graphics card in your browser for a few hours. All you need is a Google account. For this model: runs for free – The free Tesla T4 GPU with 15 GB is enough.
- Download the notebook. Click the button and save the file.
- Open Colab. Go to colab.research.google.com and sign in with your Google account.
- Upload the notebook. In the window that appears, click Upload on the left (or choose File → Upload notebook at the top) and select the file
colab.ipynb. - Turn on the GPU. In the top menu, go to Runtime → Change runtime type → under Hardware accelerator, select T4 GPU → Save.
- Start everything. Runtime → Run all (keyboard shortcut Ctrl + F9). If you see the warning “This notebook was not authored by Google,” click Run anyway.
- Wait. The cells run one after another. A green checkmark appears to the left of each cell when it's done. Installation and download take about 4 minutes here.
- Try it out. The result appears below the last cell. To test something else, change the text in the cell and run it again with Shift + Enter. The last cell starts a proper chat with an input field.
- Clean up. When you're done: Runtime → Disconnect and delete runtime. This saves your free quota.
Good to know: The free GPU is limited (usually a few hours per day). After about 90 minutes of inactivity, Colab disconnects, and you'll have to start everything over.
Windows 10 / 11
The easiest way is Ollama, a free program that launches AI models with a single command.
- Update your graphics driver (NVIDIA cards only): Install the NVIDIA app or the latest driver. For AMD cards, use the latest Adrenalin driver.
- Install Ollama. Download the file OllamaSetup.exe from ollama.com/download/windows, double-click it, and choose Install. Afterward, a small llama icon appears in the taskbar at the bottom right.
- Open a terminal window. Press the Windows key, type PowerShell and press Enter. A blue or black window opens.
- Start the model. Copy this command ("Copy" button), paste it into the window with a right-click, and press Enter:
The first time, Ollama downloads the model (1.3 GB (about 2 min at 100 Mbit/s)). After that, it's instant.ollama run embeddinggemma-2:270m - Start typing. Once
>>> Send a messageappears, type your question and press Enter. Type/byeto end the chat.
Not enough memory? Pick a smaller variant from the table above, e.g. ollama run embeddinggemma-2:270m-bf16-text.
Prefer automatic? A script that picks the right variant for you
- Download both files into the same folder: try_windows.ps1 demo.py
- Open the folder in File Explorer, right-click an empty spot and choose Open in Terminal.
- Paste this command and press Enter:
powershell -ExecutionPolicy Bypass -File try_windows.ps1
If needed, the script installs Ollama via winget, measures your VRAM (video memory) and downloads the matching variant.
macOS
Macs with an Apple chip (M1 to M5) are ideal. Older Intel Macs are very slow.
- Install Ollama. Download it from ollama.com/download/mac, open the file, and drag Ollama into your Applications folder. Launch Ollama once – a llama appears in the menu bar at the top.
- Open Terminal. Press Cmd + Space, type Terminal, and press Enter.
- Start the model. Paste the command (Cmd + V) and press Enter:
The first time, the model is downloaded (1.3 GB (about 2 min at 100 Mbit/s)).ollama run embeddinggemma-2:270m - Start typing. Type your question and press Enter. Quit with
/bye.
Linux
- Check your GPU.
nvidia-smishould show your card. If it doesn't, install the driver on Ubuntu and restart:sudo ubuntu-drivers install - Install Ollama:
curl -fsSL https://ollama.com/install.sh | sh - Start the model:
Exit withollama run embeddinggemma-2:270m/bye. Useollama psto check whether the model is running on the GPU. - Or let it pick the right variant automatically:
bash try_linux.sh
Try it
Embed an image of a dog on the beach and the search query "dog walking by the sea", then compare their similarityEmbed three English paragraphs about photovoltaics and search them by similarity for "solar panels on the roof"Embed a Python function and find matching descriptions in the code repositoryIf something goes wrong
It's an embedding model: it produces vectors, not text, and can't be used as a chat model.
Use the right task prefixes for search or classification.
Re-normalize after truncating vectors.
The source doesn't say whether image, video and audio inputs are already supported in Ollama; check the documentation if needed.
“out of memory,” a crash, or your computer freezes
The model is too big for your VRAM (video memory). Pick a smaller variant from the table above and close other programs, especially games and browsers with lots of tabs.
It runs, but extremely slowly
Your CPU is probably doing the work instead of your graphics card. In Ollama, the PROCESSOR column of ollama ps shows how much runs on the GPU. If it says "100% CPU", update your graphics driver or pick a smaller variant that fits entirely into VRAM (video memory).
“ollama” or “python” is not found
Close the terminal window and open it again so it recognizes the newly installed program. For Python on Windows: reinstall it and check Add python.exe to PATH.
Windows: "Running scripts is disabled on this system"
Run the script with exactly the command above: powershell -ExecutionPolicy Bypass -File try_windows.ps1. This allows execution for this one time only.
Error message when loading the model (version, "trust_remote_code", unknown architecture)
Brand-new models often need the latest software version. With Ollama: update Ollama. With Python: pip install -U transformers in the activated environment. The Model Card usually lists the exact version.
Tokenstrom generates the scripts and notebook automatically. Please take a quick look before running them – they're plain, readable text. Mind the model's license: apache-2.0.
Ask the AI assistant
Questions about this article, installation or background? Our assistant knows this page and runs on our own machine, not in someone else's cloud.
The Assistant is an AI (Gemma 4) and can make mistakes – please check important details against the source. Questions are used only to generate an answer and are deleted after one hour. Please don't enter any personal data.
Comments
What do you think? Already tried it? Write it here, no sign-up needed.
Comments are public. Links aren't allowed, and we delete abusive posts. You can delete your own comment at any time in this browser. Learn more in the Privacy Policy.