What is GPT?

GPT (Generative Pre-trained Transformer) is a family of large language models developed by OpenAI. The architecture uses decoder-only transformers that are first pre-trained on large text corpora to predict the next token, then fine-tuned with RLHF (Reinforcement Learning from Human Feedback) for instruction following. Major versions include GPT-2 (2019), GPT-3 (2020), GPT-3.5 (2022, powering the original ChatGPT), GPT-4 (2023), and GPT-5 (2025). GPT models pioneered the scaling paradigm in AI, demonstrating that larger models trained on more data exhibit emergent capabilities. The GPT architecture has influenced virtually every subsequent LLM.