Google Unveils Gemini 1.5 Pro With 1 Million Token Context Window

Google announced Gemini 1.5 Pro, a new generation of its flagship AI model, on February 15, 2024. The most significant advancement is its expanded context window, capable of processing up to 1 million tokens. This represents a tenfold increase from the 100,000 tokens available in Gemini 1.0 Pro, enabling the model to analyze and reason over much larger amounts of information.

This enhanced context window allows Gemini 1.5 Pro to ingest and understand extensive documents, hours of video, or lengthy codebases. For instance, the model can process a 1,500-page document, a 1-hour video, or over 11 hours of audio in a single prompt. Google demonstrated this capability by having Gemini 1.5 Pro analyze a 402-page PDF document about the Apollo 11 mission, identifying specific details and answering complex questions about its content.

Beyond the expanded context window, Gemini 1.5 Pro also incorporates a Mixture-of-Experts (MoE) architecture. This design allows the model to be more efficient by activating only specific parts of its neural network for a given task, rather than engaging the entire model. Google stated this MoE approach contributes to improved performance and scalability. The model is also said to retain the strong multimodal reasoning capabilities of its predecessor, understanding and combining information from text, images, audio, and video.

Google plans to make Gemini 1.5 Pro available to developers and enterprise customers through Google AI Studio and Vertex AI starting February 15, 2024. Initially, it will be offered with a 128,000 token context window, with the full 1 million token capacity rolling out to select customers in the coming months. This release positions Gemini 1.5 Pro as a powerful tool for complex analytical tasks requiring deep comprehension of extensive data.