OpenAI launched GPT-4.5, its newest and largest model, yet the company admits it might not perform as well as the o1 or o3-mini models.
GPT-4.5 has enhanced writing capabilities, world knowledge, and what the AI company calls a “refined personality over previous models.”
OpenAI released a video demonstrating its capabilities. The featured staff members explained that they tested the new model for EQ or “emotional intelligence.”
They showcased it by asking GPT-4.5 for an angry email.
In response, the model thought they were upset, so it asked for confirmation before writing the irate message.
The AI company says they trained this new model “using new supervision techniques combined with traditional methods, like supervised fine-tuning (SFT) and reinforcement learning from human feedback (RLHF).”
Yet, the model will only be available as a research preview.
It also “won’t introduce enough new capabilities to be considered a frontier model.”
On February 27, 2025, OpenAI opened access to the document, “OpenAI GPT-4.5 System Card.”
It says GPT-4.5 “does not introduce net-new frontier capabilities compared to previous reasoning releases.”
ChatGPT Pro users can try GPT-4.5 today.
Moreover, OpenAI will roll out access to Plus and Team users next week and to Edu and Enterprise users the week after.
It has search, file and image uploads, and canvas access. However, it doesn’t support multimodal features like video, screen sharing, and Voice Mode.