Lesson 49 · 2 min
Rate limits: why AI sometimes says "try again later"
A rate limit caps how much you can use an AI service in a period of time, so one user cannot hog shared computers and costs stay under control.
Running a model costs real money in chips and electricity. To share that fairly, services set limits: a number of messages, tokens or requests per minute, hour or day.
As a chatbot user, you meet this as a message saying you have hit your limit and can continue later, or that you have been switched to a smaller model for now. Paid plans usually raise the limits; they rarely remove them.
For developers using an API, limits are counted per account or key. Send too many requests too fast and the service answers with an error instead of a reply. Good apps wait a little and retry, slowing down each time.
Limits change often, so check the service's own help page instead of trusting numbers you read online. If you rely on AI for work, plan your heavy tasks and keep a backup tool.
Do it now
Find your own limits
- 1. Open the help or plan page of the chatbot you use.
- 2. Look for what happens when you reach your usage limit.
- 3. Note when the limit resets, so a long task does not stop at a bad moment.
Quick check
Keep going
Next →What is an API?Words from this lesson
- Rate limit
- A cap on how many requests or messages you can send to an AI service in a set amount of time.
- API
- A doorway that lets one program send requests to another, for example an app asking an AI model for an answer.