GPT-4 Release

Setting the scene

After ChatGPT's surprise public success in late 2022, generative AI moved from research labs into mainstream business, education, media, and software development almost overnight. Companies raced to integrate large language models into search, office software, coding tools, and customer support. The next question was whether models could become more reliable, multimodal, and useful for complex reasoning tasks.

What happened

On March 14, 2023, OpenAI announced GPT-4, describing it as a large multimodal model that could accept image and text inputs, though initial public access focused mainly on text. It was made available through ChatGPT Plus and an API waitlist, and Microsoft integrated GPT-4 into Bing and other products. OpenAI claimed improvements on standardized tests and professional exams, while also warning about hallucinations and safety risks.

Why it still matters

GPT-4 accelerated enterprise adoption of generative AI and intensified debates over copyright, labor displacement, misinformation, education, regulation, and AI safety. It became a reference point for the rapid commercialization of frontier models. Memorable detail: OpenAI's technical report revealed far less about model size and training data than earlier AI papers, signaling a shift from open research norms toward competitive secrecy.

Background

OpenAI released GPT-4 in March 2023, presenting it as a more capable large language model and later expanding access through ChatGPT and API products.