OpenAI Presents New GPT-4.5 Model With Improved “Emotional Intelligence”

Microsoft-backed AI powerhouse OpenAI has unveiled GPT-4.5, hailing it as its most advanced and sophisticated chatbot model yet. Designed to enhance user interactions, improve intent recognition, and exhibit a higher “EQ” (emotional intelligence), the AI model is now available as a research preview for ChatGPT Pro subscribers and developers worldwide.

Building upon the foundation of GPT-4o, OpenAI has fine-tuned GPT-4.5 to be a more versatile, general-purpose model, rather than limiting it primarily to STEM-related reasoning. The model incorporates both new supervision techniques and established methods like supervised fine-tuning (SFT) and reinforcement learning from human feedback (RLHF), ensuring improved responses and adaptability.

According to OpenAI, GPT-4.5 benefits from training on smaller models, enhancing its ability to understand context, nuance, and user intent. This results in more natural, fluid interactions, making it an ideal tool for tasks such as writing, programming, designing, and problem-solving.

One of its standout features is its ability to interpret subtle cues and demonstrate a high level of emotional intelligence, setting it apart from earlier reasoning models. Unlike previous iterations, GPT-4.5 “doesn’t think before it responds,” instead focusing on real-time interpretation to provide dynamic and engaging conversations. OpenAI has hinted that the model could unlock unexpected applications as its capabilities continue to be explored.

On the safety front, OpenAI has assured that internal evaluations of GPT-4.5 indicate no significant increase in risks compared to its predecessors. The model is also designed to reduce hallucinations, ensuring greater accuracy and reliability across a broad range of topics.

GPT-4.5 is currently accessible to ChatGPT Pro users via web, mobile, and desktop through the model picker. OpenAI plans to extend access to Plus and Team users next week, followed by Enterprise and Edu users in the subsequent rollout phase.

Users will continue to enjoy features such as Search, Canvas, and file uploads with GPT-4.5. However, multimodal functionalities like Voice Mode, video, and screen sharing are not yet supported in ChatGPT.