Large Language Models are the foundation of modern AI assistants like ChatGPT, Claude, and Gemini. They work by predicting the next token in a sequence, but this simple objective — scaled to billions of parameters and trillions of training tokens — produces emergent capabilities like reasoning, coding, and creative writing.
LLMs are built on the transformer architecture, use self-attention mechanisms to process context, and are typically trained in two phases: pre-training on broad text data and fine-tuning on specific tasks or instructions.