Google CEO Sundar Pichai announced that Google DeepMind has officially released its latest open-source multimodal large model Gemma 3, which focuses on low cost and high performance, supporting text, visual reasoning, and function calling. Gemma 3 has four parameter scales: 1B, 4B, 12B, and 27B. The largest 27B model only requires one H100 GPU to run efficiently, while similar models require at least 10 times more computing power to achieve similar performance.
Gemma 3 supports 140 languages and provides 128K contextual windows, significantly improving long text processing capabilities. According to LMSYS ChatbotArena testing, its 27B model is second only to DeepSeek R1-671B and higher than well-known open-source models such as o3-mini and Llama3-405B. In addition, Google has simultaneously launched ShieldGemma 2, which focuses on image security detection and helps developers filter sensitive content. Gemma 3 is now available for download and can be used on platforms such as Google AI Studio, Hugging Face, Vertex AI, etc.
