During the event Google I/O 2024which took place this Tuesday (14), Gigante das Buscas presented new features for the chatbot Gemini 1.5 Prowhich is now accessible to AI Premium plan subscribers on Google One.

The ability of this new version of Gemini to handle an incredible context window of 2 million tokens is something that transcends the expectations of even the most demanding users. In addition, the company also presented a new version of Gemini. I’ll explain everything to you in the next paragraphs.

Read too
Google I/O: Project Astra is artificial intelligence that understands the environment through cell phone cameras
Veo is Google’s artificial intelligence for creating videos that competes with OpenAI’s Sora

Gemini 1.5 Pro: Intuitive and Feature-Packed

Since last year, the company’s focus on AI has received well-deserved attention, with significant advancements being incorporated into Gemini 1.5 Pro. This model is capable of understanding and processing up to one million tokens, providing users with advanced capabilities to analyze documents in detail, interpret images, and much more.

Back in February, Google began testing this model with developers, allowing them to explore its vast potential. And on March 13, the company took it a step further, opening its technology to a wider audience. Subscribers to the Gemini Advanced package, integrated into Google One’s AI Premium plan, now have the opportunity to enjoy all these advanced features for R$97 per month.

The Pro version of Gemini 1.5 offers an even richer and more complete AI experience. At the launch of this model, Google highlighted its impressive analysis capacity: it is possible to examine an hour of video, listen to and interpret eleven hours of audio, review 30 thousand lines of code or delve into more than 700 thousand words. Furthermore, this service has been carefully integrated with Workspace applications, which represents a considerable advantage for business productivity.

“In the very near term, Gemini Advanced will take on the role of a versatile data analyst, discovering valuable insights and autonomously generating insightful charts from spreadsheet data,” shared Sissie Hsiao, vice president and general manager Gemini Experiences and Google Assistant. “In addition, it is important to emphasize that Gemini preserves the privacy of your files, which are not used to train our machine learning models.”

According to information released by Sissie, this feature will be available in Gemini Advanced “in more than 150 countries and can be used in more than 35 different languages”, drastically expanding the reach and usefulness of this advanced tool.

Expansion to 2 million tokens and introduction of 1.5 Flash

The Gemini 1.5 Pro has everything to raise the bar even further. During the Google I/O 2024 sessions, it was announced that an upcoming update will allow the chatbot to work with an expanded context window for 2 million tokens. This functionality is currently being tested and is exclusive to developers via the API, as well as those registered on the Google Cloud waitlist.

Demis Hassapis, CEO of the Google DeepMind division, expressed his excitement about the incremental improvements: “Not only have we added to the already robust Gemini framework an increase in the context window to an astonishing 2 million tokens, we have also made significant improvements in code generation, in improving logical reasoning and planning skills, in addition to allowing more intricate dialogues and an even deeper understanding of audio and image inputs”.

For those looking for lighter and equally effective alternatives, Google presents 1.5 Flash. Developed to be fast and efficient, 1.5 Flash is the ideal solution for tasks that demand agility and need to be performed on a large scale.

Despite being a more compact option compared to the 1.5 Pro, the 1.5 Flash does not fall short in its ability to understand multiple information simultaneously, explained Hassabis. It is particularly effective for concise summaries, supporting interactive conversations, describing image and video content, extracting data from lengthy documents and complex tables, among other applications.

Source: https://www.hardware.com.br/noticias/2024-05/google-anuncia-novidades-para-a-sua-inteligencia-artificial-gemini.html

Leave a Reply

Your email address will not be published. Required fields are marked *