Generative AI & LLMs
Max tokens
A limit on how many tokens a model is allowed to generate in a response.
Example
A response stops after 300 generated tokens because the application set that maximum.
Why it matters
It helps manage response length, waiting time and usage costs.
You'll hear it
“The reply hit the length limit before it reached the conclusion.”
Can you control it?
Sometimes
Sometimes. Some products or workflows let you influence max tokens, while others keep them fixed or manage them automatically. Check the settings and documentation for the tool you are using.
Common questions
- Will the AI always use the full allowance?
- No. It may finish earlier if it reaches a natural stopping point.
- Is this a word limit?
- Not exactly. The system counts text pieces, which do not always match whole words.
- Can a low limit cut off useful information?
- Yes. A reply may end before explaining an important qualification or completing the requested format.