OpenAI
OpenAI is an AI research and product company founded in 2015. It develops ChatGPT, the GPT model families and tools for developers. Sam Altman is its CEO. The nonprofit OpenAI Foundation controls the commercial business, OpenAI Group PBC.
Company & leadership
- CEO
- Sam Altman.
- Board chair
- Bret Taylor, OpenAI Foundation.
- Original leadership
- Sam Altman and Elon Musk were co-chairs. Ilya Sutskever was research director. Greg Brockman was CTO.
- Other founding members
- Trevor Blackwell, Vicki Cheung, Andrej Karpathy, Durk Kingma, John Schulman, Pamela Vagata and Wojciech Zaremba.
How OpenAI started
OpenAI launched as a nonprofit research company. Its 2015 announcement named a founding research team and a broad group of supporters, including technology founders, AWS, Infosys and YC Research. Those supporters committed $1 billion. That figure described funding commitments, rather than money already spent or a company valuation. The announcement said only a small part would be spent in the first few years.
Who controls OpenAI?
OpenAI created a commercial subsidiary in 2019. In October 2025, it announced a new structure: the nonprofit became the OpenAI Foundation, and the commercial company became OpenAI Group PBC. The Foundation appoints the commercial company’s directors and can replace them. At the close of that transaction, the Foundation held 26% of the equity, Microsoft held about 27%, and employees and other investors held 47%. Those are transaction-date figures, not a live ownership estimate. The distinction matters: Microsoft’s investment does not give it the Foundation’s governance rights.
The API opened a second route to OpenAI models
OpenAI introduced its API on 11 June 2020. The initial service accepted text prompts and returned text completions, letting developers build their own applications around its models. It differed from a single-purpose consumer application: developers supplied the interface, instructions and surrounding workflow. This API launch preceded the public release of ChatGPT by more than two years.
When did ChatGPT launch?
ChatGPT launched on 30 November 2022 as a research preview. The original release was designed for conversation, including follow-up questions, corrections and refusals. OpenAI described human trainers writing conversations and ranking model responses as part of its training process. That launch is a product milestone; it is not OpenAI’s founding date.
Which models does OpenAI develop?
OpenAI’s catalog includes GPT language models and separate models for images, speech, transcription, embeddings and moderation. It also includes downloadable gpt-oss 120B and 20B releases. These are separate model families with different access and deployment options. The guides below describe the reviewed releases and link to their model-specific documentation.