Google releases AI Gemini 1.5 Pro, can analyze more data and inputs


,- Google announced an upgrade to the Large Language Model (LLM) Gemini 1.5 Pro. The company announced the upgrade during the Google I/O 2024 developer conference on Tuesday (14/5/2024) U.S. time.

This LLM is used to run Gemini's artificial intelligence-based chatbot.

According to Google, the Gemini 1.5 Pro can now hold up to two million tokens. This figure is more than doubled than before. When the 1.5 Pro was announced last February, this LLM was only able to receive and process one million Tokens in a single query.

Thanks to that upgrade, the Gemini 1.5 Pro can support more inputs. More specifically, those two million tokens are equivalent to about 1.4 million words, two hours of video, or 22 hours of audio.

That support allows the Gemini 1.5 Pro to analyze large files. Usually, the LLM performance that supports multiple tokens is also improved. For example, the Gemini 1.5 Pro does not easily forget the current conversation content and conversations that are deviating from the topic.

A LLM like that can also better understand the data sources taken, so it can generate responses with a richer context.

Furthermore, this tech giant claims that the Gemini 1.5 Pro is algorithmically improved. Thus, the 1.5 Pro will be better at coding, reasoning and logical planning, conversations to audio and image understanding.

In Gemini API and AI Studio, the Gemini 1.5 Pro can also consider audio, in addition to previously supported photos and videos.

Developers who can't wait to take advantage of Gemini 1.5 Pro's capabilities, can try the 1.5 Flash that's already available in a public preview.

"The Gemini Pro is aimed at much more general or complex, often multi-stage, reasoning tasks," said Josh Woodward, VP of Google Labs, one of Google's experimental AI divisions.


(Newsline Paper Teams)

Previous Post Next Post