For a deeper dive into the technical details behind these capabilities, in addition to a comprehensive overview of our approach to responsible development, check with the Gemma 3 technical report.
Rigorous safety protocols to construct Gemma 3 responsibly
We consider open models require careful risk assessment, and our approach balances innovation with safety – tailoring testing intensity to model capabilities. Gemma 3’s development included extensive data governance, alignment with our safety policies via fine-tuning and robust benchmark evaluations. While thorough testing of more capable models often informs our assessment of less capable ones, Gemma 3’s enhanced STEM performance prompted specific evaluations focused on its potential for misuse in creating harmful substances; their results indicate a low risk level.
As industry develops more powerful models, it is going to be critical to collectively develop risk-proportionate approaches to safety. We are going to proceed to learn and refine our safety practices for open models over time.
Built-in safety for image applications with ShieldGemma 2
Alongside Gemma 3, we’re also launching ShieldGemma 2, a robust 4B image safety checker built on the Gemma 3 foundation. ShieldGemma 2 provides a ready-made solution for image safety, outputting safety labels across three safety categories: dangerous content, sexually explicit and violence. Developers can further customize ShieldGemma for his or her safety needs and users. ShieldGemma 2 is open and built to present flexibility and control, leveraging the performance and efficiency of the Gemma 3 architecture to advertise responsible AI development.
Able to integrate with the tools you already use
Gemma 3 and ShieldGemma 2 integrate seamlessly into your existing workflows:
- Develop together with your favorite tools: With support for Hugging Face Transformers, Ollama, JAX, Keras, PyTorch, Google AI Edge, UnSloth, vLLM and Gemma.cpp, you have got the pliability to decide on the most effective tools in your project.
- Start experimenting in seconds: Get easy access to Gemma 3 and start constructing instantly. Explore its full potential in Google AI Studio, or download the models through Kaggle or Hugging Face.
- Customize Gemma 3 to your specific needs: Gemma 3 ships with a revamped codebase that features recipes for efficient fine-tuning and inference. Train and adapt the model using your chosen platform, like Google Colab, Vertex AI and even your gaming GPU.
- Deploy your way: Gemma 3 offers multiple deployment options, including Vertex AI, Cloud Run, the Google GenAI API, Iocal environments and other platforms, supplying you with the pliability to decide on the most effective fit in your application and infrastructure.
- Experience optimized performance on NVIDIA GPUs: NVIDIA has directly optimized Gemma 3 models to be sure that you get maximum performance on GPUs of any size, from Jetson Nano to the most recent Blackwell chips. Gemma 3 is now featured on the NVIDIA API Catalog, enabling rapid prototyping with just an API call.
- Speed up your AI development across many hardware platforms: Gemma 3 can also be optimized for Google Cloud TPUs and integrates with AMD GPUs via the open-source ROCm™ stack. For CPU execution, Gemma.cpp offers a direct solution.
A “Gemmaverse” of models and tools
The Gemmaverse is an enormous ecosystem of community-created Gemma models and tools, able to power and encourage your innovation. For instance, AI Singapore’s SEA-LION v3 breaks down language barriers and fosters communication across Southeast Asia; INSAIT’s BgGPT is a pioneering Bulgarian-first large language model that demonstrates the ability of Gemma to support diverse languages; and Nexa AI’s OmniAudio showcases the potential of on-device AI, bringing advanced audio processing capabilities to on a regular basis devices.
To further promote academic research breakthroughs, we’re launching the Gemma 3 Academic Program. Academic researchers can apply for Google Cloud credits (price $10,000 per award) to speed up their Gemma 3-based research. The appliance form opens today, and can remain open for 4 weeks. Apply on our website.
Start with Gemma 3
As a part of our ongoing commitment to democratizing access to high-quality AI, Gemma 3 represents the subsequent step. Able to explore Gemma 3? Here’s where to begin:
Quick exploration:
- Try Gemma 3 at full precision directly in your browser – no setup needed – with Google AI Studio.
- Get an API key directly from Google AI Studio and use Gemma 3 with the Google GenAI SDK.
Customize and construct:
- Download Gemma 3 models from Hugging Face, Ollama, or Kaggle.
- Easily fine-tune and adapt the model to your unique requirements with Hugging Face’s Transformers library, or your chosen development environment.
Deploy and scale:
