Gemma 3 (12B)

Gemma 3 (12B) Description

Gemma 3 (12B) is a state-of-the-art multimodal large language model that integrates advanced text and vision processing capabilities. It features an extended 131,000-token context window, enabling deep understanding and generation over extensive inputs. Designed for versatile deployment, it balances high performance with efficiency across a wide range of devices, from mobile phones to high-end workstations.

Technical Specifications

• Context Window: 131K tokens
• Modalities: Text and Vision input processing
• Multilingual Support: Over 140 languages
• Deployment: Optimized for resource-constrained hardware including single-GPU systems

Key Features

• Multimodal Processing: Effectively handles and generates content from both textual and visual data, providing a comprehensive understanding for diverse applications.
• Extended Context Handling: Supports very long documents or multimodal inputs with 131K tokens of context.
• Global Language Coverage: Broad multilingual abilities for translation, communication, and content generation across 140+ languages.
• Efficient and Scalable Deployment: Suitable for a variety of environments, ensuring efficient operation even on devices with limited computational resources​

Usage

Ethical Guidelines

Google commits to ethical AI development by maintaining transparency about Gemma 3’s capabilities and limitations. The company advocates for responsible use of the model to minimize risks of misuse or unintended harmful outputs.

Licensing

Gemma 3 is offered under the Gemma Terms of Use, granting a commercially-friendly license that supports both research and commercial applications, while ensuring alignment with ethical standards.