Officially Launching Gemini - Google's Biggest AI Ever

On the night of December 6 (Vietnam time), Google officially introduced Gemini - the company's largest and most optimized Artificial Intelligence model, which is considered capable of competing with Open AI's GPT-4.

"Gemini is capable of complex multimodal reasoning. With a flexible design, Gemini is optimized in three different sizes: Ultra, Pro, and Nano to work on everything from data centers to mobile devices," said Sissie Hsiao, Google's vice president of products, Bard (Google's new conversational AI tool launched in March 2023).

Sissie Hsiao revealed that Gemini is coming to Bard in its biggest upgrade yet. Gemini will be rolled out to Bard in two phases: "Starting today, Bard will use a specially tuned version of Gemini Pro in English for advanced reasoning, planning, comprehension, and more. Early next year, we'll introduce Bard Advanced, which will give you first access to our most advanced models, starting with Gemini Ultra."

Sissie Hsiao also said that before the public launch, the company ran Gemini Pro through a number of industry-standard benchmarks. In 6 out of 8 benchmarks, Gemini Pro outperformed GPT 3.5, including the Massive Multitask Language Understanding (MMLU) benchmark, which is one of the leading benchmarks for measuring large AI models, and the GSM8K benchmark, which measures math reasoning at the school level.

Google also tweaked Gemini Pro in Bard to be better at understanding, summarizing, reasoning, programming, and planning. “We’re seeing great results: In anonymous reviews with third-party expert reviewers, Bard is now the most popular free chatbot compared to leading alternatives,” said Sissie Hsiao.

Users can start testing the Gemini Pro version of Bard today with text prompts, with support for other methods coming soon. It will be available in English in over 170 countries and regions, with more languages ​​and locations, such as Europe, to be added in the near future.

Ms. Sissie Hsiao also introduced Gemini Ultra - the company's largest and most advanced model. Gemini Ultra is designed for highly complex tasks and is built to quickly recognize and process different types of information including text, images, audio, video and encryption.

“One of the first ways you can try Gemini Ultra is through Bard Advanced. This is a new, advanced AI experience within Bard that gives you access to the best models we can. We are currently completing large-scale safety testing and will launch a trustless testing program before launching Bard Advanced to more people early next year,” she shared.

Google and Alphabet CEO Sundar Pichai said AI has the potential to create opportunities – from the ordinary to the extraordinary – for people everywhere. It will bring new waves of innovation and economic progress while boosting knowledge, learning, creativity and productivity at a scale we have never seen before.

The CEO also confirmed that the company is taking the next step on its journey with Gemini. This new era of models represents one of the biggest scientific and engineering efforts the world's technology "giant" is undertaking.

Gemini is a multimodal AI model developed by Google DeepMind, first mentioned in May at the Google I/O 2023 event. This is considered Google's "most serious effort" in the generative AI craze.

This is an ambitious project for Google and works similarly to the GPT-4 model currently available on ChatGPT. With Gemini, Google is one step closer to making Bard the world's most powerful AI partner.