OpenAI today held its first developer conference in its history. The artificial intelligence giant also made major announcements about its products during the event. One of these was the new version of the GPT language model (LLM) that makes ChatGPT possible.

The new version was called GPT 4 Turbo. The company described the turbo version of the GPT 4 as ‘more powerful and cheaper’.

It will be able to respond to more complex and longer commands and provide more up-to-date information

According to the statements, GPT 4 Turbo comes with many improvements such as better information about the world and longer commands. He says the model can see a lot more data with a 128k token context window. We should point out that this number is four times higher than GPT 4 and higher than all models currently on the market. To simplify it further, this value is 300 pages in a single job equivalent. In other words, he can easily summarize a book.

Additionally, GPT 4 Turbo was trained on much newer data. Language model according to OpenAI, With information dating back to April 2023 will be able to answer. For comparison, the information from the early version of GPT 4, released in March, was from September 2021. So there has been a serious improvement in this area.

Except these models Image and text-to-speech support made by DALL-E 3 owner. It also comes with six sounds for users to choose from. This way it is possible to hear the answers to each question in different voices. Let’s also add that the model will be three times cheaper than GPT 4 for developers.

GPT 4 Turbo is currently available in preview for developers. OpenAI says the model will be widely released in the coming weeks, but doesn’t give an exact date.