OpenAI, the parent company of ChatGPT AI models, announced the research preview of its much-anticipated model GPT-4.5, code-named Orion, on Thursday, February 27. The company has called GPT-4.5 their largest and best model for chat so far. Among the other features, the company boasts of the thoughtfulness and sensibility that the GPT-4.5 model adopts while responding to certain sensitive queries.

As per the company, this Emotional Quotient (EQ) is unique to the new AI model. The model’s higher world knowledge gives it yet another edge over the other AI models. These features, combined, make it useful for tasks like improving writing, programming, and solving practical problems, the company's blog states.

GPT-4.5: A New Change in AI Models

With the launch of the preview model, the company has challenged the claims of reaching the end of pre-training models. The company has successfully avoided the question of reaching the peak of pre-training data by focusing on increasing the amount of time and computing power that AI reasoning models use to “think” through problems and not just the data they have been fed on.

This will contribute significantly to the qualities of the response generated by the AI model.

Also Read: Say Goodbye to Coding Struggles! Google Launches AI-Powered Gemini Code Assist With Unlimited Code Completions

GPT-4.5: Fewer Hallucinations 

It exemplifies the power of scaling unsupervised learning by leveraging increased compute, expanded data, and advancements in architecture and optimization. These improvements enhance the model’s knowledge depth and understanding, resulting in fewer hallucinations and greater reliability across diverse topics.

Unsupervised learning is a machine learning technique where an algorithm analyzes unlabeled data to identify patterns and relationships without human guidance.

Also Read: Elon Musk’s XAI Unveiled Grok 3 this week - Don't Miss out on its Powerful New Features!

GPT-4.5: Future of AI Models 

The latest model is more general-purpose and inherently smarter. As GPT-4.5 and similar models advance through pre-training, they will provide a stronger foundation for reasoning and tool-using agents.

Conclusively, the model has two distinct strengths. First, it has a significant level of Emotional Quotient (EQ), and second, it hallucinates less owing to its grounded response and more strategic thinking. Through this, the company is trying to converge the ways of reasoning and pre-training to let an all-new world of AI models emerge.