OpenAI's GPT-5 Orion is under question: high cost and lack of data hinder progress

OpenAI is lagging behind on its plan to develop a new version of its flagship artificial intelligence GPT-5, currently codenamed Orion. The company has been working on the project for 18 months, but is facing difficulties due to a lack of the necessary data to achieve the desired level of intelligence, reports the Wall Street Journal.

What you should know about GPT-5 Orion

OpenAI has already completed two major training runs of the Orion model, each requiring months of data processing, but researchers have continually encountered new problems and the results have not met expectations. While the current version of Orion outperforms previous OpenAI models, it is still not ready for the high costs of maintaining it, as the computational resources to train such models can cost hundreds of millions of dollars.

When OpenAI introduced ChatGPT two years ago, it seemed that artificial intelligence would change our lives. However, competitors are already closing in, and OpenAI’s developments in particular relate to the ongoing struggle with a lack of data. The company is trying to improve the model by creating new data with the help of special people who solve mathematical problems and provide explanations for their actions.

The training process is very resource-intensive, and while Orion requires a significant amount of data, traditional sources do not provide sufficient results. In particular, OpenAI is testing new approaches to improve the model’s performance by developing synthetic data, creating new types of data generation, and incorporating new reasoning methods to enhance the model’s ability to solve complex problems.

OpenAI is currently aiming to launch Orion at scale, but data limitations and other factors have delayed the development process. The model is still in the development stage, and the cost of training it and integrating it into real-world environments remain significant challenges for the company.