The newest model in Google’s Gemini artificial intelligence model series, the Gemini 1.5 Flash, was introduced on Tuesday at Google I/O.
The newest model in Google’s Gemini artificial intelligence model series, the Gemini 1.5 Flash, was introduced on Tuesday at Google I/O, the company’s annual developer conference. The corporation made a point of saying that this new approach is designed to be quicker and more effective, meeting the demands of developers for both economy and speed.
Google’s CEO of DeepMind, Demis Hassabis, highlighted the advancements of Gemini 1.5 Flash during a press briefing, stating, “We heard from developers that they wanted something faster and even more cost-effective.” The new model boasts capabilities such as summarizing conversations, captioning images and videos, and extracting data from extensive documents and tables.
The launch of Gemini 1.5 Flash comes at a time when tech companies are increasingly prioritizing generative AI in their product development. This shift is crucial for Google, as generative AI provides users with more advanced and creative ways to access online information compared to traditional web search.
Meanwhile, OpenAI has also been active in the AI space, launching a new AI model and desktop version of ChatGPT on Monday. The new model, GPT-4o, promises to be twice as fast as GPT-4 Turbo while costing half as much. This competition underscores the rapid advancements and intense race in AI development.
In addition to Gemini 1.5 Flash, Google has introduced the enhanced Gemini 1.5 Pro model. This upgraded model can manage multiple large documents totaling up to 1,500 pages or summarize up to 100 emails, according to a Google vice president involved with the Gemini project. Sissie Hsiao, another vice president and the general manager for Gemini experiences, noted that the 1.5 Pro model would soon be capable of handling an hour of video content or codebases with over 30,000 lines.
“You can quickly get answers and insights about dense documents, like figuring out the details of the pet policy in your rental agreement or comparing key arguments of multiple long research papers,” Hsiao explained.
OpenAI’s recent upgrade enhances ChatGPT with improved quality and speed, supporting 50 different languages. This new model will be accessible through OpenAI’s application programming interface (API), allowing developers to build applications using the upgraded model immediately.
On the other hand, Google claims that Gemini 1.5 Pro, which supports 35 languages, features a 2 million token window—a measure of context that indicates how much information the model can process at once. Company executives also mentioned improvements in local reasoning, planning, and image understanding.
Alphabet CEO Sundar Pichai highlighted the practical applications of this advancement by giving an example of a parent using Gemini to summarize all recent emails from their child’s school. “It offers the longest context window of any foundational model yet,” Pichai stated during the press briefing.
Initially, Gemini 1.5 Pro will be available for testing in Workspace Labs. Gemini 1.5 Flash will also be accessible for testing and on Vertex AI, Google’s machine learning platform that enables developers to train and deploy AI applications. This phased rollout aims to ensure robust performance and user feedback before wider release.