Google Unveils Gemma 3: A Powerful Multimodal, Multilingual Open-Source LLM with Long Context

Google has officially released Gemma 3, its latest open-source large language model, boasting multimodal capabilities, multilingual support, and an extended context window. Designed to compete with other open models, Gemma 3 can process text, images, and video, and handles over 100 languages. The model's long context length—up to 128K tokens—enables it to analyze entire documents or long conversations in one pass. Available in several sizes, Gemma 3 aims to democratize access to advanced AI for developers and researchers, offering free use under a permissive license. Early benchmarks show strong performance on reasoning, summarization, and multilingual tasks.