On December 6, 2023, Google launched Gemini, describing it as “our most capable and general model yet, with state-of-the-art performance across many leading benchmarks.” The announcement, featuring Google and Alphabet CEO Sundar Pichai and Google DeepMind CEO Demis Hassabis, framed Gemini as “built from the ground up to be multimodal, which means it can generalize and seamlessly understand, operate across and combine different types of information.” Google said Gemini Ultra was “the first model to outperform human experts on MMLU (massive multitask language understanding).”
Gemini arrived in three sizes for different uses: Ultra (the largest and most capable), Pro (for a wide range of tasks), and Nano (for on-device use). The accompanying technical report, “Gemini: A Family of Highly Capable Multimodal Models” from the Gemini Team at Google, was submitted to arXiv on December 19, 2023, detailing models that process text, images, audio, and video.
Gemini mattered as the first flagship to come out of the combined Google DeepMind - the merger of Google Brain and DeepMind earlier in 2023 - and as Google’s direct frontier-model response to GPT-4. It established the Gemini line that continued through later generations, including Gemini 3 in 2025.