Google has launched Gemini, its next-generation large language model that will power the likes of Google Bard chatbot and the company’s Pixel 8 Pro smartphone. The search giant is calling Gemini its “most capable and general model," and in being so—replaces PaLM 2. It has been built to be multimodal—meaning that it can understand and operate using different forms of information such as text, code, imagery, and even video.
“This new era of models represents one of the biggest science and engineering efforts we’ve undertaken as a company. I’m genuinely excited for what’s ahead, and for the opportunities Gemini will unlock for people everywhere," Sundar Pichai said at the launch.
Gemini Is Available In Three Sizes To Suit Different Platforms
Apart from setting new records in benchmarks, Gemini is also flexible. Google says it can run on data centres; at the same time, it’s designed to be able to run on something as pocketable as a mobile phone.
Google Gemini 1.0 is available as Gemini Ultra, Gemini Pro (powers Google Bard), and Gemini Nano (for phones like Pixel 8 Pro). This is expected to help developers and enterprise-grade customers to easily scale and build using AI.
Beats OpenAI’s GPT-4 In Benchmarks
We also have a direct head-to-head comparison with GPT-4, which powers ChatGPT Plus and Microsoft Bing AI Copilot. As per Google, Gemini outperforms GPT-4 in Text-based benchmarks in general, reasoning, math, and code categories, and replicates the same in multimodal benchmarks in image, video, and audio categories.
Moreover, Gemini Ultra achieved a score of 59.4% on the new MMMU benchmark, and it “consists of multimodal tasks spanning different domains requiring deliberate reasoning," Google said. It also outperforms previous models in image benchmarks without help from Object Character Recognition Systems (OCR).
Gemini Is Already Rolling Out Across Google Products
To experience Gemini’s capabilities, you can use Google Bard to see how Gemini Pro performs compared to PaLM 2. Google is calling this the biggest update to Bard since it launched earlier this year.
In addition, Google Pixel 8 Pro users would also be able to use Gemini Nano through apps like Recorder, Gboard, and even WhatsApp, with support for more chat apps coming later.
Also, Google says that Gemini would be available in services like search, Ads, Chrome, and Duet AI soon. And, for developers and enterprises, the company will make Gemini Pro available through the Gemini API in Google AI Studio or Google Cloud Vertex AI.
Google Bard ‘Advanced’ Coming Soon
Gemini Ultra will power Google Bard Advanced starting in early 2024, and Google is also taking a more responsible approach towards the Ultra variant by completing trust and safety checks before making it widely available. It, too, will be available to select developers, customers, and partners and “safety and responsibility experts" in 2024.
Comments
0 comment