v2.2 · 2026-06-12

EmbeddingGemma 300M

embeddinggemma-300m

EMBEDDING

EmbeddingGemma 300M is a 0.308B embedding model profile for local AI planning. This page records provider, license, context, quantization, hardware-fit estimates, setup hints, source links, and caveats so readers do not choose a model by name alone.

PROVIDER

Google DeepMind

FAMILY

Gemma / EmbeddingGemma

MODEL TYPE

embedding

PARAMETERS

0.308B

MODALITIES

text

ARCHITECTURE

—

CONTEXT WINDOW

2,048 tokens

TRAINING TOKENS

—

RELEASE DATE

2025-09-04

LICENSE & LINKS

Gemmalicense text

MODEL CARD

huggingface.co/google/embeddinggemma-300m

HUGGING FACE

google/embeddinggemma-300m

SETUP HINTS

OLLAMA

ollama pull embeddinggemma

LM STUDIO

Available according to the supplied June 2026 research; select a cited GGUF/MLX build and verify current app metadata before relying on hands-on setup guidance.

LLAMA.CPP

Use a cited GGUF build with current llama.cpp-compatible tooling.

RAM / VRAM ESTIMATES

MIN RAM

0.2 GB

MIN VRAM

0.2 GB

These are conservative local-inference estimates, not official hardware requirements.

HARDWARE FIT

CPU only (no GPU)

limited

8 GB RAM

limited

16 GB RAM

limited

32 GB RAM

limited

8 GB VRAM

limited

12 GB VRAM

limited

24 GB VRAM

limited

Apple Silicon (unified memory)

limited

Hardware fit values are conservative local-inference estimates based on GGUF size plus runtime and KV-cache overhead. Actual requirements depend on context length, quantization, runtime, GPU offload, and other running apps.

BEST FOR

·Private offline search, mobile RAG, and simple document lookup on laptops without dedicated GPUs locally.

AVOID IF

·Your documents exceed 2K tokens per chunk, or you need multimodal retrieval support today locally.

CAVEATS

·Google documentation uses 308M parameters, while some catalog surfaces use the 300M label.

·The context window is 2K tokens, so long documents should be chunked carefully.

·The model uses the Gemma license, not Apache-2.0; users must accept and comply with Google terms.

·Google states sub-200MB RAM with quantization, but exact Q4, Q5, and Q8 file sizes were not confirmed in cited sources.

·Commercial-use status from the supplied research: Allowed under Gemma license and Google terms; responsible commercial use is permitted with prohibited-use restrictions.

·Beginner summary from supplied research: EmbeddingGemma is Google’s small text embedding model for private local search. It turns notes, documents, or app text into vectors on laptops and mobile-class devices, making offline RAG practical without heavy GPU hardware for beginners.

SOURCE URLS