Gemini Ultra and ChatGPT 4: A Side-by-Side Comparison

Author(s): Roman Orac

Originally published on Towards AI.

There’s a lot of buzz about Gemini Ultra. See how it actually compares to ChatGPT in my latest deep dive.
Gemini Ultra and ChatGPT 4: A Side-by-Side Comparison (image by DALL-E 3)

Google recently announced their new large multi-modal model called Gemini Ultra. They’ve highlighted the model performance in this video.

It’s an inspiring video because it shows the future of computer and human interaction, basically how we might use this technology in the future.

OpenAI has a similar model called GPT-4 Vision, which was released in September 2023. I wanted to see how GPT-4 Vision would perform given the same tasks as Gemini did.

So, I’ve made screenshots of the Gemini video and asked GPT-4 Vision to explain them.

Comparing Gemini Ultra responses with ChatGPT (video by author).

Let’s start with a multi-modal dialogue test. I’m using a simple prompt: What do you see?

A sticky note (screenshot by author).

Both ChatGPT and Gemini were able to correctly identify the object in the image as a sticky note. ChatGPT labeled it directly as a “sticky note” while Gemini provided the more generic description of “a piece of paper on the table”.

The presenter in the Gemini video incrementally sketched out a duck in the water, adding more details in each step. At each stage, Gemini provided remarks on the drawing’s progress.

I simulated this with screenshots, which ChatGPT examined… Read the full blog for free on Medium.

Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.

Published via Towards AI