Shortly after launching Gemini, a large language model that Google hopes to take to the top of the artificial intelligence industry, the company is already announcing its successor. Today Google is launching Gemini 1.5 and making it available to developers and enterprise users before fully rolling out to consumers. The company positions Gemini as a business tool, a personal assistant, and a tool for everything in between.
The Gemini 1.5 Pro, Google’s general-purpose model, is obviously on par with the Gemini Ultra that the company recently released, and beats the Gemini 1.0 Pro by 87% in tests. The model was created using a technique known as Mixture of Experts, or MoE. This means that it only runs part of the overall model when you submit a request, rather than processing the request with its full capacity all the time. This approach should make the model faster to use and more efficient for Google to operate.
Gemini 1.5 has a huge context window, which means the model can handle much larger queries and view much more information at once. This window is a whopping 1 million tokens, compared to 128,000 in GPT-4 and 32,000 in the current Gemini Pro. Google CEO Sundar Pichai explains it: “It’s about 10 or 11 hours of video, tens of thousands of lines of code.” The size of the context window means you can ask the AI bot about all of this content at once. Pichai also said that Google researchers are testing a context window with 10 million tokens – that’s like all the episodes of Game of Thrones at once.
Sundar Pichai also believes that a large contextual window will be very useful for business. “This is useful in cases where you can add a lot of personal context and information at the time of the request.” For example, filmmakers can upload their entire film and ask Gemini what reviewers say. Companies will be able to use Gemini to view masses of financial records at once. “I consider this one of the biggest breakthroughs we’ve made.”
For now, Gemini 1.5 will only be available to business users and developers through Google’s Vertex AI and AI Studio. It will eventually replace Gemini 1.0, and the standard version of Gemini Pro available to everyone on gemini.google.com and the company’s apps will be 1.5 Pro with a contextual window of 128,000 markers. You’ll have to pay extra to get a million. Google is also reviewing the model’s safety and ethical boundaries, particularly regarding the new larger context window.
Google is now in a race to create the best artificial intelligence tool as companies around the world try to determine their own artificial intelligence strategy: whether to strike deals with OpenAI, Google or someone else. Just this week, OpenAI announced memory for ChatGPT and is looking to roll out its own web search. So far, Gemini seems amazing, especially for those already in the Google ecosystem, but there is still a lot of work ahead on all sides.
Over time, all these versions 1.0 or 1.5, pluses and ultras will not matter to users. “People will just consume the experience. It’s like using a smartphone without paying attention to the processor in it,” says Sundar Pichai. But right now, he says, we’re still at the point where everyone knows the chip in their phone because it matters.
Source: The Verge
Join the ITS Authors Contest! Win cool prizes from our partners Logitech – professional gaming steering wheel and low-profile gaming keyboards.