What is an APC?
Agent Processing Credits (APCs), also known as tokens, are the basic unit of work that an AI model reads and writes. Within Lyzr we use the term APC for differentiation from raw LLM token pricing — but technically, 1 APC = 1 token.
What is a token?
A token is a chunk of text that an AI model processes. It can be a full word, part of a word, a number, punctuation, or even a space. As a rough rule for English text:
- 1 token ≈ 4 characters
- 100 tokens ≈ 75 words
- 1–2 sentences ≈ 30 tokens
What counts toward APC usage
APC usage is not just the visible answer. Every model call adds up:
System prompt, user message, history, tool definitions, retrieved docs, files, records.
The visible response generated by the model.
Hidden model computation used by reasoning-capable models, when billed.
Tool schemas, tool call results, shell output, file reads, search results, handoffs.
A simple chat = one call. An agent run = many calls. A multi-agent workflow = dozens. That's why transcripts look small while real consumption is much larger.