OpenAI today announced the launch of its new GPT-4.5 model, which it described as the largest and most known model to date, but it is not a Frontier Model, and may not outperform other company models such as the O1 or O3-mini.

The GPT-4.5 model features improved writing abilities, a wider knowledge of the world, and a more sophisticated personality than previous versions. However, the model does not offer enough technical leaps to rank in the leading category, according to the company.

OpenAI explained that the new model improves GPT-4's computational efficiency by more than 10 times, but does not add substantial new capabilities in the field of inference compared to previous versions, and its performance is less than O1 and O3-mini in most tests.

According to the company's early tests, users found that the new model offers a more natural dialogue experience, with the ability to express warmth, and show a level of emotional intelligence that was not available in previous versions.

OpenAI attributes the notable improvements in GPT-4.5 to advances in unsupervised learning, providing the model with an unclassified dataset, allowing it to extract patterns and insights from it itself.