Token
The chunk of text a language model actually reads and writes — roughly three quarters of an English word on average; generating might split into gener + ating. Models have token limits (context windows), and API pricing is usually per token, so token counts translate directly to cost. On VAR2, tokens are also the platform currency for generations — same word, related idea: every generation consumes compute that is measured and priced in tokens. Practical tip: long prompts are fine for image models, but with LLMs, concise prompts save money and leave room for output.