Despite the many companies that are getting involved with their own projects in the artificial intelligence sector, it could be said that OpenAI is one of the main references. All this apart from other technological giants such as Microsoft or Google.
And many of these companies want to make a place for themselves in this growing AI sector that doesn’t seem to have an end. With everything and with this and despite the large number of platforms of this type that we can use, perhaps the best known is ChatGPT. Furthermore, this is a chatbot that has not stopped evolving over the months due to the different language models it uses.
These models are becoming more advanced and focus, in some cases, on certain modes of work and behavior. What there is no doubt about is that, for some time now, one of the most anticipated elements of this type is GPT-5. Here we refer to the advanced language model which has been talked about for so long and which will represent an enormous advance in the world of artificial intelligence.
The problem comes in that there is already speculation with the launch of GPT-5 for too long now at the hands of the aforementioned OpenAI. The main reason for this delay is that, from what we know now, the company is having serious problems in its development and training. It is important to keep in mind that before the implementation of a new intelligent language model, it needs a intensive training by its creators.
Reason why GPT-5 is so late
And that is precisely where the company is encountering crucial obstacles that are significantly delaying the launch of the new model. In fact, this is not something that happens for the first time, since something similar happened with the current GPT-4 that was released last year in 2023.
To give you a closer idea of what is happening, let’s say that the lack of data for the language training It is the main obstacle that the development of GPT-5 is facing. As we are now learning, AI pre-training is reaching a critical and problematic point, all due to the scarcity of data available online.
Therefore, the lack of data for training the language model is currently leading OpenAI to try to find alternative methods to achieve its goal. For example, you are starting to use and generate synthetic data for all of this. In parallel, the company is acquiring the services of experts in certain specific areas so that GPT-5 can continue to evolve before its official staging.
To all this we must add the huge economic inversion that is being carried out to move the project forward. Each GPT-5 training cycle is around 500 million dollars. Hence the company is focusing on more affordable models before the arrival of its long-awaited GPT-5.