OpenAI
What
AI research company founded 2015. Creator of GPT series, o1/o3 reasoning models, Codex, DALL-E, Sora. Became industry leader by revenue and arguably capability through 2025. Structurally transformed from non-profit to capped-profit to for-profit 2024-2025.
Key Contributions Relevant to This Wiki
- o1/o3 models: Demonstrated test-time compute via chain-of-thought reasoning at scale; o1-preview was watershed for reasoning model mainstream awareness
- Harness engineering: Used Codex agents internally to ship production systems via environment design rather than direct coding
- GPT-4 class models: Established the benchmark-beating paradigm that commoditized inference
- RLHF: Pioneered reinforcement learning from human feedback as alignment technique (now widely adopted)
Key People (Historical)
- Sam Altman (CEO)
- John Schulman (co-founder, left 2024) — see Lilian Weng acknowledgments
- Lilian Weng (former Head of Safety)
- Andrej Karpathy (former, now independent)