OpenAI Releases GPT-4

Summary

OpenAI released GPT-4, its most capable large language model to date, marking a significant leap in multimodal AI capabilities. The model accepted both text and image inputs and demonstrated human-level performance on a range of professional and academic benchmarks, including passing the bar exam in the 90th percentile.

What Happened

On March 14, 2023, OpenAI publicly launched GPT-4, the successor to GPT-3.5 which had powered the original ChatGPT. GPT-4 was a large multimodal model capable of processing both text and image inputs and generating text outputs. OpenAI described it as "the latest milestone in its effort to scale up deep learning."

The model demonstrated substantial improvements over its predecessor across virtually every benchmark. It scored in the 90th percentile on the Uniform Bar Exam (compared to GPT-3.5's 10th percentile), passed the Biology Olympiad in the 99th percentile, and showed marked improvements in factual accuracy and reasoning tasks.

OpenAI released a technical report but notably withheld details about model architecture, training data, and compute used — a stark departure from the organization's founding ethos of openness. The company cited both competitive pressures and safety concerns as reasons for this opacity.

GPT-4 launched initially as part of ChatGPT Plus ($20/month) and through an API with a waitlist. Microsoft, which had invested $10 billion in OpenAI earlier that year, integrated GPT-4 into Bing Chat.

Why It Matters

GPT-4's release marked several inflection points simultaneously. As a technical achievement, it demonstrated that scaling continued to yield substantial capability gains, fueling the "scaling laws" hypothesis that more compute and data reliably produce better models. As a commercial product, it cemented the subscription-based access model that would come to define the industry.

The decision to withhold architectural details was arguably as significant as the model itself. It crystallized the tension between OpenAI's original mission of open research and its evolution into a competitive commercial enterprise. This opacity became the template for subsequent frontier model releases and intensified the open-vs-closed debate that would dominate AI discourse throughout 2023 and 2024.

GPT-4 also reset public expectations for what AI could do, triggering a wave of enterprise adoption and government attention that culminated in the Biden Executive Order on AI later that year.