TL;DR: Every leap in LLMs, from GPT-1 to today's agents, came from changing one thing: what the model is trained to do. Next-token prediction gave models knowledge. Post-training made them assistants. Reinforcement learning made them reason. Tools made them agents. Underneath it all, the real engine is data designed to teach the right objective. Every few weeks, a new LLM launches and the feed calls it a revolution. Taken one at a time, the launches look like noise. There is a simpler way to read the story. Each big step answered one question: what should we train the model to do? First the answer was "predict the next word." Then it became "be a helpful assistant," then "think before you answer," and now "take actions until the job is done." Each step kept everything before it and added one idea on top. This mental model helps me make sense of the field, and it explains both how we got here and where things are going. Act 1: Predict the next token (2018 to 2020) GPT-1 (2018): the objective works. GPT-1 was a 117M-parameter transformer trained on one task: predict the next token in about 7,000 books. To do that well, it had to pick up grammar, facts, and meaning on its own. Fine-tuned on small labeled datasets, it beat purpose-built models on 9 of 12 benchmarks. Next-token prediction turned out to be a way to learn language itself. GPT-2 (2019): scale, and tasks start to appear. GPT-2 grew to 1.5B parameters trained on 40 GB of web text. Without fine-tuning, it began to summarize, translate, and answer questions zero-shot. The results were rough, but they hinted that scale alone unlocks new abilities. GPT-3 (2020): no more fine-tuning. At 175B parameters, GPT-3 could learn a task from a description and a