OpenAI introduced the GPT-4 language model, which can handle text and images

The notorious ChatGPT chatbot runs on GPT-3.5. The day before, OpenAI announced a new language model. It has several advantages and is called GPT-4.

Here's What We Know

The developers claim that GPT-4 is the world's first experience with a high-performance artificial intelligence system. The company notes that it will be difficult for users to notice a difference when communicating with the chatbot, but in more complex tasks the differences will become apparent.

According to OpenAI, the GPT-4 language model didn't just manage to pass the bar exam. Its result is in the top 10. The developers claim that the new model proved to be more creative than the current version. In particular, it can write scripts and even songs.

GPT-4 is able to generate responses consisting of 25,000 words. The limit has been increased by more than 4 times compared to ChatGPT. Another important innovation is that the new model can analyse images. It will only respond in text form.

OpenAI is already working with companies that are ready to integrate the new language model into their products (Duolingo, Stripe and Khan Academy). GPT-4 will be available as an API and for a $20/month ChatGPT Plus subscription.

Source: OpenAI