株式会社オブライト
AI Model2026-05-17

Gemma (Google)

Also known as: Gemma (Google) / Google Gemma / Gemma 4

Google's open-weights LLM family — distilled from Gemini-class technology and shipped in local-friendly E2B / E4B / 26B / 31B variants.


Overview

The Gemma 4 family includes both MoE and dense variants runnable on consumer GPUs. The edge-focused E4B packs 4.5 B parameters with multimodal support. See Gemma 4 hardware requirements.

Licence and use cases

Distributed under the commercial-friendly Gemma licence and easily deployed via Ollama or HuggingFace, standing alongside Llama and Qwen for on-prem AI and edge inference.

Related Columns

Related Terms

Feel free to contact us

Contact Us