All tasks / Models
EmbeddingGemma 2
Google's small open-weights multilingual text embedding model, built from Gemma and sized for on-device use.
- Open weights
- Released Sep 2026
Where it ranks
- –Embeddings & retrievalModels · Local · no result on this task’s benchmarks yet
Facts
- Vendor
- Family
- EmbeddingGemma
- Released
- Sep 2026
- Weights
- Open: download and run it yourself
- License
- Apache-2.0
- Parameters
- 0.74B
- GPU memory to run locally
- ~2 GB (Derived: 0.74B parameters x 2 bytes = ~1.5 GB of bf16 weights.)
- Access
- Free (open weights)
- Last verified
- 19 Oct 2026
Benchmark results
No public benchmark results recorded yet.
Where you can use it
- SOllamaVector databases & engines · Local
- SText Embeddings InferenceVector databases & engines · Local
- SvLLMVector databases & engines · Local
Earlier versions
- EmbeddingGemma 300M · Sep 2025