Gemma
Google DeepMind's family of open-weight foundation models — derived from the same research as Gemini, available in sizes from 2B to 27B for local and cloud deployment
Free to download and run; available via Google AI Studio, Vertex AI, and Hugging Face
Overview
Gemma is Google's family of lightweight, open-weight language models built on the same research and technology as Gemini. Designed to run efficiently on consumer hardware — including laptops, phones, and edge devices — Gemma models are popular for local AI applications, fine-tuning experiments, and privacy-sensitive deployments.
Key Features
- Open weights with permissive license for research and commercial use
- Model sizes from 2B to 27B for different hardware capability levels
- Gemma 3: latest iteration with improved instruction following and reasoning
- ShieldGemma: safety-focused variant for content moderation
- CodeGemma: fine-tuned for code completion and generation tasks
- Optimized for efficient inference on CPUs, GPUs, and mobile devices
Pricing: Free and open-source; run locally or via Google Cloud Vertex AI.
Pros
- Derived from Gemini research — one of the strongest open model foundations available
- Multiple specialized variants: CodeGemma, PaliGemma, instruction-tuned
- Runs on consumer hardware — deployable locally without cloud infrastructure
- Permissive terms with no commercial use restrictions
Cons
- Significant capability gap compared to closed Gemini Pro and Ultra
- Fine-tuning and local deployment require ML infrastructure knowledge
- Llama 3 has a larger community and more fine-tuned variants available
Tags
Product Updates
Similar Tools
Phi
Microsoft's family of small but exceptionally capable open-weight language models that outperform models many times their size
Black Forest Labs
The AI lab behind FLUX — the leading open-source image generation model family that set a new standard for photorealism and prompt accuracy
Falcon
Open-source foundation models from the Technology Innovation Institute in Abu Dhabi — among the first truly open, commercially licensed large language models
Hugging Face
The leading open-source hub for AI models, datasets, and demos — home to hundreds of thousands of models for every AI task
