Large Language Models (LLMs) are AI systems that read, understand, and generate human language. They are built with transformers and trained on large text corpora. This training lets them learn grammar, facts, and simple reasoning patterns. LLMs turn input text into numerical units called tokens. The model processes these tokens internally and then generates one token at a time to form a response. The output tokens are converted back into readable text. A tokenizer performs this mapping. It splits text into words, subwords, or characters and assigns numerical values the model can process. LLMs are often trained as foundational models. They are not designed for one fixed task but can adapt to many. After broad training, they can be fine-tuned or guided with examples for tasks such as question answering, summarization, or story writing. State-of-the-art LLMs are called frontier models. They define current AI capability but require high computational and energy resources to train.