Gemini 2.5 Pro ships with a 1 million token context window, soon expanding to 2 million tokens. This enables it to process vast datasets across text, audio, images, video, and code repositories, making it one of the most versatile AI models available.