Google revealed a new model of artificial intelligence called Gemini 1.5 Flash, which focuses on efficiency and speed, during the developer conference Google I/O 2024.
Dennis Hassabis, CEO of Google’s DeepMind AI subsidiary, stated that Gemini 1.5 Flash excels in a variety of tasks such as summarization, chat, image and video commenting, and data extraction from documents.
Hassabis explained that Google launched the Gemini 1.5 Flash version to meet developers’ needs for a lighter and less expensive model than the previously announced Gemini 1.5 Pro in February, surpassing the capabilities of the original Gemini version launched last year.
According to Google, the new model – despite being lighter than the Pro version – surpasses the Gemini 1.5 Nano designed to work on devices.
Google mentioned that they maintained the performance of the new Flash version through a process called “distillation,” transferring knowledge and core skills from the previous version to the current one. This allowed them to continue supporting multimedia and a long context window that helps the model handle a large amount of data simultaneously, with its capacity reaching up to a million unique tokens.
Google stated that the Gemini 1.5 Flash model will be able to analyze a document containing 1500 pages or a software database of over 30,000 lines instantly.
It should be noted that the Gemini 1.5 Flash model is not intended for consumers but for developers, providing a cost-effective and faster way to build AI-based products and services using Google’s technologies.
In addition to introducing the Flash model, Google is working on updating the Gemini 1.5 Pro model to enhance its capabilities in language processing, voice, and image analysis. It is expected that Google will expand the current context window of the model to reach two million special tokens later this year, allowing it to process massive amounts of data.
Google now offers the Gemini 1.5 Flash and Gemini 1.5 Pro models in public preview mode through the AI Studio and Vertex AI services.
While these updates are not aimed at the average user, they provide exciting opportunities for developers to create advanced applications and services based on artificial intelligence.
