On November 30, 2022, OpenAI released ChatGPT to the public as a free research preview. The company described it simply as “a model called ChatGPT which interacts in a conversational way.” Within five days, ChatGPT reached one million users. It became the fastest-growing consumer application in history and catalyzed what would become known as the AI boom.

What it was

ChatGPT was a conversational interface built on GPT-3.5, a variant of OpenAI’s GPT-3 model fine-tuned using Reinforcement Learning from Human Feedback (RLHF). The RLHF process trained the model to follow instructions, maintain dialogue context, refuse harmful requests, and produce more helpful and less toxic outputs than the base model.

Unlike the GPT-3 API, which required users to carefully craft prompts for a text-completion system, ChatGPT presented a chat interface where users could simply ask questions and have back-and-forth conversations. This lowered the barrier dramatically. Users who had never interacted with a large language model could now hold extended conversations, ask follow-up questions, and request revisions.

The free research preview had no usage limits at launch. Users could generate poems, debug code, explain concepts, draft emails, and experiment with creative writing, all without paying or applying for API access.

What made it different

ChatGPT was not the first large language model or the first chatbot. What made it different was the combination of:

  1. Capability: GPT-3.5 was substantially more capable than prior publicly accessible models at following complex instructions and maintaining coherent multi-turn conversations.

  2. Accessibility: The chat interface required no technical knowledge. Users typed in natural language and received natural language responses.

  3. Free unlimited access: The research preview had no paywall and no rate limits at launch, allowing viral experimentation.

  4. RLHF alignment: The model had been trained to be helpful, harmless, and honest (within limits), making it more useful for real tasks than a raw completion model.

The result was an explosion of use cases. Within days, social media was flooded with examples: students using it for homework, developers generating code, writers brainstorming plots, marketers drafting copy. The viral loop was self-reinforcing because each creative use case inspired others to try their own experiments.

The numbers

ChatGPT reached one million users in five days. For comparison:

  • Netflix took 3.5 years to reach one million users
  • Facebook took 10 months
  • Instagram took 2.5 months
  • Spotify took 5 months

By January 2023, ChatGPT had over 100 million monthly active users, making it the fastest-growing consumer application ever recorded at the time. OpenAI reported that by February 2026, ChatGPT had crossed 900 million weekly active users.

What happened next

The ChatGPT launch triggered a cascade of competitive responses and investment:

  • Microsoft announced a $10 billion investment in OpenAI in January 2023 and began integrating GPT into Bing and Microsoft 365.

  • Google declared a “code red” and accelerated its own AI efforts, leading to the rushed launch of Bard in February 2023.

  • Anthropic, founded by former OpenAI researchers, raised hundreds of millions to build Claude.

  • Meta released LLaMA in February 2023, opening the door to open-source competition.

  • Venture funding for AI startups exploded in 2023 and 2024.

OpenAI introduced ChatGPT Plus ($20/month) in February 2023, offering priority access and faster responses. GPT-4 followed in March 2023, bringing multimodal capabilities and improved reasoning. Enterprise offerings, plugins, and the GPT Store followed as OpenAI built a product ecosystem around the conversational interface.

Why it still matters

ChatGPT did not invent large language models or conversational AI. What it did was prove that LLMs were ready for mass-market consumer use and that the interface for AI could be as simple as a text box.

The ChatGPT launch established several patterns that shaped everything that followed:

  • Chat as the default interface: Every major AI lab now ships a chat product.
  • RLHF as standard practice: Human feedback training became the default for production LLMs.
  • Free tiers as growth engines: Anthropic, Google, and others followed with free access to drive adoption.
  • AI as infrastructure: The assumption that AI should be integrated into every product became mainstream.

For builders, the lesson is that capability matters, but access matters more. GPT-3 was powerful but obscure. ChatGPT was powerful and available. The difference in impact was enormous.

Sources

Further reading