Volume
/ day
%
Embed step (per request)
tok
Retrieve step
chunks
tok
Generate step
tok
tok
Evaluate step (judge)
tok
Estimated monthly cost
$3937 / mo
300,000 requests / month — 3,900 prompt tokens each
| Step | Model | Monthly |
|---|---|---|
| Embed | OpenAI text-embedding-3-small | $4.80 |
| Generate (input) | GPT-4o | $2048 |
| Generate (output) | GPT-4o | $1800 |
| Evaluate | GPT-4o mini | $84.38 |