Google has once again pushed the boundaries of artificial intelligence with the introduction of Gemini 1.5, the latest iteration of its groundbreaking Gemini series. This next-generation model is designed to offer users even greater contextual understanding and more helpful features than its predecessor.
In only one week following the launch of Gemini Ultra 1.0, Google has made significant strides and unveiled the next-generation Gemini 1.5. The Gemini 1.5 Pro is the first model of the Gemini 1.5 that Google has made available for testing. Google CEO Sundar Pichai said in a blog post that "it shows dramatic improvements across a number of dimensions and 1.5 Pro achieves comparable quality to 1.0 Ultra, while using less compute."
Pichai claims that by enabling Gemini 1.5 to comprehend a large amount of data at once, Google has improved its intelligence. It can now process one million bits of data at once, which is more than any large model has ever been able to achieve. "This indicates that 1.5 Pro can handle enormous volumes of data at once, such as over 30,000 lines of code or over 700,000 words in codebases, as well as one hour of video and eleven hours of audio. Up to 10 million tokens have been successfully tested as part of our study," claims Google.
The new Mixture-of-Experts (MoE) architecture used by the Gemini 1.5 is clever, according to Demis Hassabis, CEO of Google DeepMind, since it only employs the portions of the model's brains required for the task at hand. This improves and speeds up its operation.
In December, we launched Gemini 1.0 Pro. Today, we're introducing Gemini 1.5 Pro! 🚀
This next-gen model uses a Mixture-of-Experts (MoE) approach for more efficient training & higher-quality responses. Gemini 1.5 Pro, our mid-sized model, will soon come standard with a… pic.twitter.com/m2BNufHd8C
Intriguingly, Google tested the Gemini 1.5 Pro model with a 44-minute silent film, and the results showed that the model could "accurately analyse various plot points and events, and even reason about small details in the movie that could easily be missed."
To show what’s possible with the drastically huge context window in Gemini 1.5 Pro, we prompted it with the three.js examples code - over 100,000 lines of code/800k+ tokens!
(That’s not even the max, it can handle millions of tokens 😀)
Gemini was able to process all the code… pic.twitter.com/N2VYgn2JFJ
Gemini 1.5 is currently exclusively accessible to developers and business users. Google has not yet disclosed the date or method of its public release. Google has reaffirmed that it is "committed to bringing each new generation of Gemini models to billions of people, developers, and enterprises around the world responsibly" in its announcement blog.