OpenAI Chief Executive Officer (CEO) Sam Altman on Friday announced the launch of GPT - 4.5, the company's most advanced model yet. The new AI model is designed to be more general purpose, according to Altman.
In a post on X, Altman praised the new model as “talking to a thoughtful person.” He said the model's intelligence had left him “astonished” at several moments, and he promised users an experience that felt like "magic".
“This isn't a reasoning model and won't crush benchmarks. It's a different kind of intelligence…really excited for people to try it!” the CEO said.
While hailing the model, Altman also acknowledged that it was expensive and of massive scale. The initial rollout was intended to include both Plus and Pro tiers simultaneously. However, due to rapid growth, the company faced a GPU shortage. For this, OpenAI plans to add new GPUs next week.
GPT-4.5 is ready!
February 27, 2025
good news: it is the first model that feels like talking to a thoughtful person to me. i have had several moments where i've sat back in my chair and been astonished at getting actually good advice from an AI.
bad news: it is a giant, expensive model. we…How Is GPT 4.5 Different from Earlier Models?
According to OpenAI, GPT 4.5 is its most advanced model, designed to be more general-purpose compared to its reasoning-focused models. It was trained using new supervision techniques combined with reinforcement learning from human feedback. GPT-4o was also trained similarly.
OpenAI said that the model feels more “natural.” It has a stronger alignment with user intent and improved emotional intelligence. It is expected to produce improved content for writing, programming and solving practical problems.
With a broader knowledge base, the new model is also expected to experience fewer hallucinations. It has been designed to recognise patterns, draw connections, and generate creative insights without reasoning, OpenAI added.
Unlike o-series models, GPT 4.5 would not produce detailed step-by-step logic, thereby making it more natural. OpenAI also told the users that the new model would not be able to produce audio or video and could be slower due to its size. However, GPT-4.5 could support image inputs.
The US-based tech giant further explained that the model has gone through extensive safety evaluations and poses no enhanced safety risk compared to its existing models.
How To Try GPT 4.5
The model will be available for OpenAI's ‘pro' users globally. This means that paying users can access the model on their web, mobile, and desktop.
Essential Business Intelligence, Continuous LIVE TV, Sharp Market Insights, Practical Personal Finance Advice and Latest Stories — On NDTV Profit.
