Understanding the language of single-cell biology
Based on Gemma and developed in collaboration with Yale University, Cell2Sentence-Scale 27B helped discover a potential new pathway for cancer therapy.
Welcome to the Gemmaverse
Explore Gemma models crafted by the community
Slide 1 of 3
Towards interspecies communication
Researchers at the Wild Dolphin Project and Georgia Tech treat dolphin vocalizations as a learnable language that can be tokenized, analyzed, and generated—just like human speech—with DolphinGemma.
Developing for Indic languages
Despite the incredible advances of AI, challenges remain in ensuring equitable representation and performance across diverse languages. Watch how developers at Telugu-LLM-Labs used Gemma to create Navarasa.
Enabling sovereign AI
Gemma open models give governments and enterprises a base to build powerful sovereign AI systems, with complete control over their data and infrastructure
Slide 1 of 3
The Ministry of Economy, Ecology and Agriculture of Ukraine digitizes licensing process with Gemma
INSAIT creates leading Bulgarian-first LLM with Gemma 2
Introducing GAIA, a Brazilian Portuguese Gemma 3 model developed with ABRIA, CEIA, Nama, and Amadeus AI
The Ministry of Economy, Ecology and Agriculture of Ukraine digitizes licensing process with Gemma
INSAIT creates leading Bulgarian-first LLM with Gemma 2
Introducing GAIA, a Brazilian Portuguese Gemma 3 model developed with ABRIA, CEIA, Nama, and Amadeus AI
Enterprises and startups building with Gemma
Explore more open models in the Gemmaverse
After fine-tuning with Gemma, the model outperformed other SOTA models across internal benchmarks all while running faster and achieving a 98% cost reduction for our fintech agent.
Build responsibly
Open models and tools are developed in accordance with AI Principles
Slide 1 of 3
ShieldGemma 2
A defensive filter that sits between the user and the model to intercept and block toxic inputs or outputs
VaultGemma
Trained with a differential privacy guarantee to ensure it learns new concepts without ever memorizing or leaking sensitive information
Gemma Scope
A microscope for the model's inner workings using sparse autoencoders to visualize exactly how and why it makes specific decisions