GPT-3 (Generative Pre-trained Transformer 3) and GPT-4 (Generative Pre-trained Transformer 4) are both language models developed by OpenAI.
GPT-3 was an advanced version in the GPT series, released in June 2020. It had 175 billion parameters, making it a large language model. GPT-3 was capable of generating highly coherent and human-like text, completing tasks such as language translation, summarization, and question-answering with impressive accuracy. It was used in various applications such as chatbots, content generation, and even writing entire news articles.
GPT-4 was just released hours ago.
Soon, you should be able to use 4 to process images to turn them into text. For example, you can give it an image and then tell it to write a blog post from it. Or, you could show it a meme and ask why it is funny and what the context is. However, these features are not released widely – yet.
GPT-4 can find loopholes in contracts and develop advantages for the user, plus it appears to be able to find answers and help with taxes. Although it is recommended that one still employs an accountant to look over the results.
It still does not seem to have access to the internet – but this is probably a blessing in disguise for now.
In order to access most of these new features, it seems that someone needs to have the paid-for account with ChatGPT plus OR they can muddle around using Bing chat.