Tokens explain why LLM prompts cost what they cost

This article explains that an LLM does not read text as normal letters. It first breaks text into tokens, and each model may split the same sentence differently. That matters for AI agents because more tokens can mean slower runs and higher API cost.

Key points

  • More tokens usually mean more work for the model and higher API cost.
  • Non-English text can use more tokens depending on the model and tokenizer.
  • BPE is a common method that turns frequent text pieces into tokens.
  • Different tokenizers can count the same sentence in different ways.

Quick term guide

FIR
A First Information Report — the official complaint filed with police in India that kicks off a criminal investigation.
tokens
Tokens are small pieces of text that AI systems count when reading or writing.
token
A small piece of text used to measure AI input, output, and cost.
diff
A view that shows exactly what changed in the code.
AI agents
AI agents are AI tools that can carry out steps toward a goal, not just answer once.
AI agent
An AI program that can inspect information and suggest what to do next.
agents
AI helpers that follow your instructions and make changes for you.
tokenizer
The tool that splits text into tokens before the model sees it.
Read original