What is a token, and why does the same prompt cost different amounts on different models?

A token is the atomic unit an LLM reads and generates — roughly a common character sequence like a word, subword, or punctuation. It's not a word: "GPT" is one token, " GPT" (with a leading space) is another, and "aiengineering" is usually several. Each provider trains its own tokenizer (Anthropic, OpenAI, Google all differ) so the same English sentence encodes to different token counts across models — that, plus different per-token prices, is why the same prompt costs different amounts.