Tokens, context window and model size

Learn the words you will see everywhere: tokens (pieces of words; about 4 characters or ¾ of an English word; Hindi and Hinglish use more tokens), context window (how many tokens the model can read at once — prompt + documents + answer), parameters (model size, e.g. 1B, 8B, 70B) and why bigger models are smarter but slower and costlier. Cost is charged per input and output token.