Published on May 14, 2024, 2:31 pm

Google recently introduced an upgraded version of its generative AI model, Gemini, at the Google I/O 2024 developer conference. The new Gemini 1.5 Pro boasts enhanced capabilities, now able to analyze longer documents, codebases, videos, and audio recordings with up to 2 million tokens — effectively doubling its previous capacity limit.

The increased token input of the new Gemini model sets it apart in the AI landscape, allowing for more comprehensive analysis equivalent to around 1.4 million words, two hours of video content, or 22 hours of audio recordings. Models with larger token inputs like Gemini 1.5 Pro are less prone to forgetting recent information and can provide contextually rich responses by grasping the flow of data more efficiently.

Developers keen on exploring this advanced AI technology can sign up for the private preview of Gemini 1.5 Pro with a 2-million-token context via Google AI Studio. This offering comes alongside Gemini 1.5 Flash, a streamlined version optimized for speed and efficiency in tasks such as summarization, chat applications, and image or video captioning.

Google is also rolling out features like context caching and Batch API within its Vertex AI platform to enhance cost-effectiveness for developers utilizing Gemini models. Additionally, controlled generation capability is set to debut soon in Vertex AI, enabling users to define desired output formats and potentially reduce overall expenses associated with utilizing extensive context.

The company’s relentless commitment to advancing its generative AI technologies reflects Google’s ambition to empower developers worldwide with cutting-edge tools for various applications across different industries. Stay tuned as Google continues to innovate and push boundaries in the field of artificial intelligence and generative models.


Comments are closed.