At 10 requests/user/day (totaling 15,000,000 queries/mo), running RAG / search-augmented answers on GPT-4o mini costs $18,000.00 / month ($0.36 / user / mo).
Total expected API invoice for 50,000 Active Users generating 15,000,000 queries.
Direct inference cost per MAU to model your SaaS pricing tiers and gross margins.
Savings generated by exploiting 50% cache hits on prompt context.
| Model | Provider | Monthly Spend (50,000 Active Users) | Cost / User / Mo | Annual Spend | Context Limit |
|---|---|---|---|---|---|
| GPT-4o mini (Current) | openai | $18,000.00 | $0.36 | $216,000.00 | 128,000 |
| GPT-5.6 Luna | openai | $22,200.00 | $0.444 | $266,400.00 | 1,050,000 |
| GPT-5.4 mini | openai | $83,250.00 | $1.665 | $999,000.00 | 256,000 |
| GPT-5.4 nano | openai | $22,575.00 | $0.4515 | $270,900.00 | 128,000 |
| Claude Haiku 4.5 | anthropic | $103,500.00 | $2.07 | $1,242,000.00 | 1,000,000 |
| Claude 3.5 Haiku | anthropic | $82,800.00 | $1.656 | $993,600.00 | 200,000 |
At 10 queries per user/day with GPT-4o mini, the estimated cost is $0.36 per monthly active user (MAU). Total monthly bill for 50,000 Active Users is $18,000.00.
Assuming a 50% cache hit rate on repeated system and context tokens, prompt caching saves $4,500.00 per month ($54,000.00/year) on GPT-4o mini.
With a direct COGS of $0.36 per user/month on GPT-4o mini, charging at least $1.80 per user/month ensures an 80%+ software gross margin.