Google introduced Gemini in 2023 as a new multimodal AI model family and used it to replace the LaMDA and PaLM-based technology behind its Bard chatbot.
Google and DeepMind announced Gemini in December 2023. The model was designed to handle multiple kinds of information, including text, images, audio, video, and code. Google released versions with different sizes and capabilities, including Gemini Ultra, Gemini Pro, and Gemini Nano.
Bard was renamed Gemini in February 2024, making the model’s name also the name of Google’s consumer chatbot and assistant experience. This change connected Google’s public AI product more directly with the underlying model family.
Gemini is not the same as BERT, T5, or MUM. BERT was introduced for bidirectional language representation, T5 framed many language tasks as text-to-text problems, and MUM was designed for complex search-related understanding. Gemini’s defining launch emphasis was multimodal capability and a unified model family.