OpenAI Launches GPT-4.5, Its Next General-Purpose Large Language Model

On Thursday, OpenAI announced the research preview of its latest large language model, GPT-4.5. This new model will primarily be available to software developers and users with ChatGPT Pro subscriptions.

According to OpenAI, the model is designed to provide more accurate information compared to earlier versions. In a blog post, OpenAI highlighted that early testing indicates that interacting with GPT-4.5 feels more natural. The new model benefits from a broader knowledge base, an improved ability to understand user intent, and enhanced emotional intelligence. These improvements make it especially useful for a range of tasks, including writing enhancement, programming assistance, and solving practical problems. OpenAI further noted that GPT-4.5 exhibits a lower rate of hallucinations compared to its predecessor models, GPT-4o and o1.

Despite these advancements, OpenAI's CEO and co-founder, Sam Altman, mentioned on social media that GPT-4.5 will not be the leading model in benchmark tests, yet he described it as having a unique quality that he has not experienced before.

This launch is significant as OpenAI is considered one of the most valuable startups globally. Last month, CNBC reported that the company was discussing a major funding round worth $40 billion at a valuation of $340 billion.

Launch Details and Access

OpenAI plans to roll out GPT-4.5 to users with ChatGPT Plus and Team subscriptions in the upcoming week, as stated by Alex Paino, who is part of OpenAI's technical team. Following that, ChatGPT Edu and Enterprise subscribers will gain access in the following week.

In a post, Altman described GPT-4.5 as a "giant, expensive model" and mentioned that the company initially intended to launch it simultaneously for Plus and Pro users. However, due to rapid growth and a shortage of GPUs, they will expand GPU availability significantly next week.

Moreover, GPT-4.5 has already become available through Microsoft’s Azure AI Foundry. Microsoft has invested over $13 billion in OpenAI and integrates its models into various products, also providing the necessary computing resources.

The release comes in a period of market speculation concerning other AI developments, particularly following significant moves by a competitor, the Chinese lab DeepSeek. Around the same time, Nvidia’s market capitalization dropped sharply, losing nearly $600 billion in a day, an indicator of how competitive the AI landscape has become.

Future Developments

A couple of weeks later, Altman stated the company's intention to improve public communication regarding their roadmap. He revealed that GPT-5 would be the next release, which will incorporate a wider range of OpenAI's technologies. Additionally, while GPT-4.5 will be the last of the non-chain-of-thought models, the upcoming versions will utilize reasoning models that apply multiple calculations in real-time as users submit their queries.

OpenAI, GPT-4.5, AI