Gemma is Google DeepMind's collection of lightweight, state-of-the-art open models built from the same technology that powers Gemini models. The platform offers specialized variants for medical, coding, and research applications including MedGemma, CodeGemma, and PaliGemma. It provides developer-friendly sizes optimized for various devices from mobile to enterprise deployments. Gemma models support wide language capabilities and integrate with popular frameworks like Hugging Face, PyTorch, and JAX.

• Multiple specialized model variants
• Multimodal capabilities
• Mobile and edge optimization
• Wide language support
• Framework compatibility
• Research and commercial use

• AI researchers
• Developers
• Academic institutions
• Enterprise AI teams

• AI research and development
• Custom model deployment
• Medical AI applications
• Code generation tools

• Hugging Face Transformers
• PyTorch and JAX
• Google Cloud integration
• MediaPipe support

https://ai.google.dev/gemma/responsible

• Model setup takes 1-2 hours for basic deployment, enterprise integration within 1-2 weeks.

• Hugging Face Models
• Ollama
• ONNX Runtime

• Open source - Free to use
• Commercial licensing available
• Cloud deployment costs separate