Gemma 4

Gemma 4 is a family of advanced open models designed to deliver high intelligence-per-parameter for developers building production-grade AI systems. Based on Gemini 3 research and technology, Gemma 4 supports complex reasoning, multimodal understanding, and multilingual interaction, making it suitable for a wide range of applications, from conversational agents to specialized enterprise tools.

Key capabilities include native support for agentic workflows, allowing developers to create autonomous agents that can plan, call functions, interact with external tools and APIs, and complete multi-step tasks reliably. Gemma 4 offers strong multimodal reasoning across text, images, and audio, enabling use cases such as document understanding, visual analysis, and voice-based interfaces. With support for 140 languages, the models can handle not only translation but also culturally aware content generation and understanding. Fine-tuning options allow teams to adapt Gemma 4 to domain-specific tasks, improving accuracy and relevance using common ML frameworks and techniques.