Gemma 3

Gemma 3 is a family of open AI models from Google DeepMind, built on Gemini 2.0 technology and available in a chat-style interface.

What Gemma 3 can do

Gemma 3 comes in four sizes—1B, 4B, 12B, and 27B parameters—so you can pick a model that fits your hardware and tasks. It supports text, images, and short videos, and can handle long prompts.

• Generate and rewrite text in 140 languages (including Russian)
• Analyze images and short videos
• Use function calling for more complex workflows
• Work with up to 128K tokens of context
• Write code in multiple programming languages
• Perform math and logical reasoning

Google notes that Gemma 3 27B can deliver performance comparable to much larger models (for example, Llama-405B) when running on a single GPU.

How to access and download

You can use Gemma 3 in Google AI Studio via the web, or download the model for local use and API integration.

• Open Google AI Studio
• Sign in with a Google account
• Select a Gemma 3 model (1B–27B)
• Adjust settings (language, functions)
• Enter a prompt and review the output

Gemma 3 is free to use. If the service is unavailable in certain regions (including Russia), you may need a workaround for regional restrictions. Downloads are available on Hugging Face, Ollama, and Kaggle.