Understanding tokens and credits
Tokens measure what models consume. Credits are how you pay for it. Here is how the two relate.
When teams first budget for AI features, two terms tend to get blurred: tokens and credits. They describe different things, and keeping them separate makes cost planning far easier.
What a token is
Language models process text as tokens — short chunks that are often a word or part of a word. Every request has input tokens (your prompt, system instructions, and any context) and output tokens (what the model generates). Providers count both.
Token counts depend on the tokenizer each model uses, so the same sentence can produce slightly different counts across model families. Non-English text and code can also tokenize differently from English prose.
What a credit is
Credits are a platform balance. Rather than paying for every request individually, you hold a balance that usage is drawn from. Credits simplify procurement and give teams a predictable monthly envelope.
Why the distinction matters
- Token counts tell you how efficient your prompts are.
- Credit balances tell you how much budget remains.
- Model-specific rates connect the two.
If costs rise unexpectedly, check token counts first: longer prompts, verbose outputs, or retrieval that adds too much context are common causes. Then review whether the chosen model is the right fit for the task.