Meet Gemini: Google’s Largest and Most Powerful AI Model | by Natassha Selvaraj | Dec, 2023


This next-gen AI model outperformed ChatGPT on almost all academic benchmarks.

Photo by Mitchell Luo on Unsplash

When OpenAI released ChatGPT last November, there was a pressing question on everyone’s minds — what are the tech giants doing?

When will companies like Google respond to this development?

We now have our answer.

On December 6, 2023, Google announced their latest AI model: Gemini.

According to the company’s CEO Sundar Pichai, this technology is a huge leap forward in artificial intelligence, and will affect virtually all of Google’s products.

The current version of the model is called Gemini 1.0. It can work with text, images, videos, and audio, and comes in 3 different sizes.

Gemini Nano is a small, more efficient version that can be run natively and on Android devices.

Gemini Pro, the middle tier version, strikes a balance between between capability and efficiency. This model performs significantly better than Google’s previous flagship model, PaLM-2. It currently powers the Bard chatbot.

Finally, Gemini Ultra is the most powerful model in the series. It excels at complex reasoning, and has outperformed OpenAI’s GPT-4 model on various benchmarks.

Gemini Ultra hasn’t been made publicly available just yet. Google has announced that the model will be launched early next year, although no specific time-frame has been provided.

OpenAI’s ChatGPT is currently powered by 2 AI models — GPT-3.5 for the free version and GPT-4 for the paid version.

A few months ago, OpenAI announced that GPT-4 had multimodal capabilities (i.e. it was able to process text, audio, and images).

However, although it can process various data types, the model’s primary design and functionality is focused on text-based inputs and outputs.

This means that GPT-4 is a text-based framework, and vision and…



Source link

Leave a Comment