6 interesting things ChatGPT 4o can do

OpenAI recently released its next flagship model GPT-4o and showed off some interesting demos. Human-like voice chat has become a standout feature, but it does much more than that. OpenAI doesn't highlight the many cool things ChatGPT 4o can do. Let's learn about the exciting new capabilities of ChatGPT 4o through the following article!

1. Create precise text in images

Diffusion models have difficulty generating text on images. Dall -E 3 still fails to create an image with the given text. However, the ChatGPT 4o model is an end-to-end multimodal model that can display text accurately. OpenAI did not mention this in the presentation. However, an example can be found on OpenAI's site where the company explores the model's capabilities.

It can create and add text to images easily. The consistency across multiple samples is remarkable. You can also attach images and ask to create images from different angles of the same character, and ChatGPT 4o maintains consistency in all situations. It can also create 3D views of objects, which can be combined to create 3D renderings. Not to mention ChatGPT 4o can also create fonts.

Keep in mind that these capabilities are not yet available on ChatGPT. It still uses Dall -E 3 to create images. OpenAI may unlock these features in the near future.

2. GPT-4o can also process video

OpenAI does not mention that GPT-4o can also process video. On the model page, OpenAI demonstrated that you can upload a video and ask GPT-4o to summarize it. From transcription to bulleted summaries, ChatGPT 4o does it all. So it seems that the Gemini 1.5 Pro is not the only model that can handle video.

3. GPT-4o can be your tutor

During a presentation with Khan Academy's Sal Khan, OpenAI showed off an engaging demo using the GPT-4o model. Basically, on iPad you can share your screen with ChatGPT 4o and it can see everything on your screen.

Now, you can ask it to explain and help find a solution to a problem. Be it math, science, charts, maps or anything else, ChatGPT 4o will be your personal teacher to guide you throughout the lesson. It's a fantastic application of AI, powered by the multi-modal vision capabilities of the GPT-4o. By the way, it also works with the ChatGPT desktop application for macOS.

4. ChatGPT 4o can be your meeting companion

In one of the demos, OpenAI introduced that users can use ChatGPT 4o as a live companion during meetings. You can share your screen with ChatGPT 4o so it can see and hear all participants. It can also provide input and participants can also ask questions to the GPT-4o model. ChatGPT 4o responds naturally and continues to participate in the conversation. Finally, you can ask it to summarize the meeting. Isn't it amazing?

5. Improve non-English language performance

OpenAI not only improves GPT-4o's performance in English, but also improves performance in other languages. It has significantly improved the model's ability to compress non-English languages ​​to accommodate more tokens.

To give some examples, Gujarati language takes up 4.4 times less tokens, 2.9 times less Hindi tokens, 3.5 times less Telugu tokens, 2.5 times less Urdu tokens, Russian tokens are 1.7 times less, etc. Basically, for languages ​​other than English, ChatGPT 4o becomes even more powerful.

6. ChatGPT 4o beats all other AI models

OpenAI doesn't discuss standard numbers and focuses on delivering new experiences. However, ChatGPT 4o overshadows all other AI models from Google, Anthropic, Meta, etc. In fact, it performs better than OpenAI's own GPT-4 Turbo model released a few months ago. before.

From MMLU to HumanEval, GPQA and DROP, ChatGPT 4o outperforms both proprietary and open source models. In the LMSYS arena too, the ChatGPT 4o model achieved an overall ELO score of 1310, much higher than other AI models.

Micah Soto

Update 20 May 2024