Gemini: Google’s Response to GPT-4

Google has unveiled Gemini, their largest and most capable AI model, surpassing OpenAI’s GPT-4. Gemini is a multimodal model pre-trained on audio, image, and text data from the beginning, scoring 90% on MMLU. It comes in three sizes: Gemini Ultra, Pro, and Nano, with Nano already running on Google Pixel 8 Pro. Gemini will power Google Bard and a more advanced version is planned for next year.