PROMPTS & APIs • 100 / 397
Understand how applications send instructions to models and handle production API behavior.
TPM
TPM means tokens per minute, a rate limit based on total model token processing.
Think of it like
Think of TPM like a restaurant or service desk: the request must reach the right endpoint in the right format and the response must be handled safely.
Real life
Web applications use concepts such as TPM to communicate with external services.
SRE lens
A few huge prompts can exhaust TPM before RPM.
Remember thisTPM means tokens per minute, a rate limit based on total model token processing.