Completion
The model’s output — the text it generates to continue or answer your prompt.
Think of it like
The reply to your letter: you wrote the prompt, this is what comes back.
Example
Send “Write a haiku about rain” and the three-line poem you get is the completion.
How it actually works
Completion is what the model produces token by token in response to the prompt. The name comes from the original framing — the model “completes” the text you started. Its length counts against your context budget and drives most of your latency.
For product teams
You pay for completion tokens too, and they usually dominate latency — long answers cost more twice over.
For engineers
The generated continuation; the decoded token sequence following the prompt.
Related
- Prompt — Responds to a prompt.
- Autoregressive — Built one token at a time via autoregressive.
Read anything AI without the jargon
Look up any term in plain English, or save terms as you read with the free Chrome extension.
Open DecoderAdd to Chrome