Skip to main content
OpenAI was founded in December 2015 by Sam Altman, Elon Musk, Greg Brockman, Ilya Sutskever, Wojciech Zaremba, and other leading technologists. Their mission: to research and develop artificial general intelligence (AGI) in a way that is safe, ethical, and benefits all of humanity. From the outset, OpenAI has focused on automation, creativity, decision-making, and problem-solving using state-of-the-art machine learning techniques.
Learn more about AGI and its implications in the OpenAI Charter.

​
Key Milestones

YearModelDescription
2019GPT-2A large-scale transformer that generates coherent, contextually relevant text.
2020GPT-3Expanded to 175 billion parameters, powering chatbots, content creation, and more.
2021CodexTranslates natural language into code, enabling developers to build applications by describing their intent.
2021DALL·EGenerates high-quality images from text prompts, unlocking new creative workflows in design and digital art.
2023GPT-4Multimodal model with enhanced language understanding, coding capabilities, and image inputs.

​
2019 – GPT-2

In early 2019, OpenAI unveiled GPT-2, showcasing how transformer architectures can produce long-form, context-aware text. Despite its capabilities, the full model was initially withheld due to concerns about potential misuse.
  • Transformer-based architecture
  • 1.5 billion parameters
  • Demonstrated coherent text generation across diverse topics

​
2020 – GPT-3

The release of GPT-3 in mid-2020 marked a major leap:
  • 175 billion parameters for deep language understanding
  • Zero- and few-shot learning capabilities
  • Applications: chatbots, virtual assistants, automated content writing, and more
Read more about the capabilities of large language models in the GPT-3 paper.

​
2021 – Codex

Building on GPT-3, Codex was introduced to bridge natural language and programming. It can:
  • Interpret plain-English prompts
  • Generate code in multiple languages (e.g., Python, JavaScript)
  • Power developer tools like GitHub Copilot
Use Codex responsibly. Generated code may require security reviews and testing before production deployment.

​
2021 – DALL·E

DALL·E demonstrated creative AI by converting text descriptions into images:
  • Leveraged GPT-style transformers for image synthesis
  • Supported diverse artistic styles and novel object combinations
  • Found use cases in marketing, digital art, and rapid prototyping

​
2023 – GPT-4

The latest milestone, GPT-4, extends OpenAI’s models to be truly multimodal:
  • Processes both text and image inputs
  • Delivers finer-grained reasoning and problem-solving
  • Powers advanced applications in research, software development, and creative industries
The image is a timeline of key milestones in AI development from 2019 to 2023, highlighting the releases of GPT-2, GPT-3, Codex, DALL-E, and GPT-4, along with their capabilities.

​
Further Reading

​
References

Watch Video