ChatGPT at Four: The Launch That Rewired the Industry

Four years ago this month, OpenAI posted a short announcement titled "Introducing ChatGPT." It described a model that could answer follow-up questions, admit its mistakes, and challenge incorrect premises, offered free as a research preview while the company gathered feedback. ChatGPT was fine-tuned from a model in the GPT-3.5 series that finished training in early 2022, built with reinforcement learning from human feedback as a sibling model to InstructGPT. The framing was modest: an experiment, not a product. It turned out to be the most consequential product launch of the decade.

The launch: an interface made the model legible. Chatbots were not new in November 2022. What ChatGPT did differently was wrap a large language model in a dialogue format anyone could drive, with no prompt engineering required. Ask a follow-up and it answered; point out a mistake and it conceded; push a false premise and it pushed back. The capabilities that had lived inside APIs became visible to anyone with a browser. Practically overnight, the "research preview" was being used for homework, code fixes, recipes, and arguments.

The numbers: a growth curve no one had seen. Five days after launch, ChatGPT crossed one million users, a milestone OpenAI's chief executive noted at the time. By January 2023, a UBS analyst note drawing on Similarweb data estimated roughly 100 million monthly active users, just two months after launch, making ChatGPT the fastest-growing consumer application in history at the time. UBS analysts wrote that in twenty years of following the internet space, they could not recall a faster ramp in a consumer internet app. TikTok had needed about nine months to reach the same mark; Instagram took roughly two and a half years.

GPT-4: the moment the stakes became concrete. On March 14, 2023, OpenAI announced GPT-4, a multimodal model accepting text and image inputs. The company said the new model was more steerable than GPT-3.5 and more likely to produce factual responses, and claimed "human-level performance" on professional and academic benchmarks. GPT-4 reached the public through ChatGPT Plus subscribers first, with developers queued on an API waitlist. It was the clearest signal yet that the capability curve was not flattening.

The legacy: a race that is still running. ChatGPT turned conversational AI into a mainstream product, and the industry has compounded from there. The launch triggered the generative-AI investment wave, from venture capital to data-center buildouts to a global shortage of GPUs. It also started the race that produced today's agentic assistants, which do not just answer questions but plan and execute tasks; open-weight models that put frontier-adjacent capability in anyone's hands; and AI tools embedded in the developer workflows of nearly every software team.

What changed. Reading the announcement from 2026, the most striking shift is in the baseline. Before November 2022, a machine that could converse was a research artifact; after it, a default. The chatbot of 2022 became the assistant of 2026, and the interface moved from answering to doing. A generation of products, companies, and open-source projects now takes that step for granted. OpenAI asked the world for feedback on a research preview. Four years later, the conversation it started is still setting the pace.