Large Language Model

Roughly speaking, the core skill of a large language model is "predicting what most likely comes next, based on what came before." It read a huge amount of code and docs during training, so it can write code, explain errors, and rewrite copy. Tools like Cursor, Claude Code, and v0 are the "workbench." Behind them are large models from different vendors.

It's not a search engine or a database. Its knowledge stops at its training cutoff. It doesn't know the code in your project unless that code is put into the conversation. And it can say wrong things with total confidence. This is called Hallucination.

Different models have different strengths. Some are better at code, some are faster and cheaper. When one model doesn't do well on a problem, trying another one is a valid option.