OpenAI, the company behind the hugely popular tool ChatGPT, has announced its latest AI model, GPT-4. GPT-4 builds on the success of its predecessor, adding features such as image input, improved accuracy of results, and enhanced creativity. In this article, we take a closer look at the new features of GPT-4, compare it to his GPT-3.5, the model that powers ChatGPT, and learn how to use GPT-4 today.
What is GPT-4?
OpenAI has been releasing GPT (Generative Pre-trained Transformer) models since 2018, and GPT-4 is the latest AI language model developed by the company. Many significant improvements have been made over GPT-3.5, the model behind the very popular tool ChatGPT. system”.
The company claims that GPT-4 can solve more difficult and complex problems with greater accuracy thanks to its broader general knowledge and problem-solving abilities. Additionally, GPT-4 will be available as an API for developers to build applications and services. Many companies such as Duolingo, Be My Eyes, Stripe and Khan Academy have integrated GPT-4 into their products. The company also live-streamed a demo of his GPT-4 on YouTube to showcase its capabilities.
GPT-4: New features compared to ChatGPT
According to OpenAI, GPT-4 outperforms previous models in three main areas: creativity, image input support, and longer context. Let’s take a closer look at each of these areas.
Image input support
One of the most notable changes from the previous generation is that GPT-4 is “multimodal”. In other words, it supports multiple forms of input. ChatGPT, and all services based on GPT-3, are restricted to text input only. The new GPT-4, on the other hand, supports both text and image input, and a combination of both.
So you can ask GPT-4 to generate image captions and describe what is happening in the scene, but its understanding goes beyond that. For example, users can show the ingredients in their kitchen and ask what they can make. During the live demo, OpenAI President and Co-Founder Greg Brockman was also able to: Generate the entire HTML code for your website Based on sketch only.
Additionally, Be My Eyes, an app that connects blind and partially sighted people with sighted volunteers, has also begun integrating into GPT-4 services. In a demonstration, the service showed off the capabilities of his GPT-4, able to correctly identify plants, describe clothing patterns, translate labels, read maps, and much more. .
longer context
Another improvement over ChatGPT (or rather GPT-3) is that GPT-4 can “keep in mind” more context. If you’ve ever used ChatGPT on your iPhone or laptop, you must have noticed that you “forget” the conversation after the point. This is because the maximum capacity is 2048 tokens, which is about 1500 words.
By comparison, GPT-4 has a maximum token count of 32,768, so it can handle longer form inputs. This means you can feed up to 25,000 words to your model based on GPT-4. This is useful if you have a large document to summarize or if you are using GPT tools to create long-form content.
GPT-4 is more creative
OpenAI claims that GPT-4 has vastly improved its ability to generate creative content and collaborate with users on creative projects. This includes a variety of uses such as composing music, script writing, technical writing, and even adapting to a user’s individual writing style.
Other features
- Advanced inference capabilities: GPT-4 has a broader general knowledge and deeper understanding of different domains than ChatGPT. Passing the Unified Bar Exam and the Biology Olympiad, GPT-4 does very well.
- Multilingual: GPT-4 is multilingual and can answer multiple-choice questions with high accuracy across 26 languages, from Italian to Korean.
- Safer, more aligned: According to OpenAI, GPT-4 is more secure and more tuned. They are 82% less likely to generate negative content in response to a prompt and 40% more likely to generate necessary information.
GPT-4: Restrictions
GPT-4 is definitely a powerful language model, but it still has some limitations. OpenAI recognizes that social bias, hallucinations, and hostile prompts remain issues that need to be addressed. The company says it continues to address these issues with “transparency, user education and broader AI literacy.”
Additionally, GPT-4, like ChatGPT, still has limitations when it comes to knowledge bases. Lack of knowledge about events occurring after September 2021but still can’t access online sources.
GPT-4: How to use it now
There are currently two ways to create GPT-4 models. The best way to use GPT-4 today is to use ChatGPT Plus, a paid version of OpenAI’s ChatGPT. According to the company, ChatGPT Plus has been upgraded to his GPT-4 AI model. Access is still (somewhat) restricted, and OpenAI says it will adjust usage caps “depending on demand and system performance.” Currently, paying subscribers are limited to 100 messages every four hours on his GPT-4.
To upgrade to ChatGPT Plus, visit the chat.openai.com website.in the left menu[Plus にアップグレード]Click the button. next,[プランのアップグレード]Clicking the button will take you to the Stripe checkout page. Please fill in the required information and complete the checkout. As a paying customer, you will immediately have access to the new GPT model.
Another way to access the GPT-4 model is through Microsoft’s Bing Chat. Microsoft has confirmed that its new AI-based Bing search engine runs on GPT-4, customized for search. If you’ve used the new Bing Chat Preview in the last five weeks, you’ve already experienced an early version of this powerful model.
However, Bing Chat has some limitations. For example, it currently does not support image input, but this feature may be added at a later stage. However, it does not support multimodal input for the time being. Additionally, it should be emphasized that Microsoft uses a combination of his GPT-4 and proprietary Prometheus model to provide information and results to keep the model on track.
You can sign up for Bing Chat using the official Bing apps for iOS and Android. You can also access the service in your Edge web browser by visiting this link. Please note that there may be a waiting list, so you may not be able to use it immediately.
What do you think of GPT-4? How does it improve your workflow? Let us know in the comments section below!
