At 20 requests/user/day (totaling 6,000,000 queries/mo), running Content generation on Llama 3.2 3B Instruct costs $504.00 / month ($0.0504 / user / mo).
Total expected API invoice for 10,000 Active Users generating 6,000,000 queries.
Direct inference cost per MAU to model your SaaS pricing tiers and gross margins.
Savings generated by exploiting 30% cache hits on prompt context.
| Model | Provider | Monthly Spend (10,000 Active Users) | Cost / User / Mo | Annual Spend | Context Limit |
|---|---|---|---|---|---|
| Llama 3.2 3B Instruct (Current) | meta | $504.00 | $0.0504 | $6,048.00 | 128,000 |
| GPT-5.6 Luna | openai | $9,340.80 | $0.9341 | $112,089.60 | 1,050,000 |
| GPT-5.4 mini | openai | $35,028.00 | $3.503 | $420,336.00 | 256,000 |
| GPT-5.4 nano | openai | $9,700.80 | $0.9701 | $116,409.60 | 128,000 |
| GPT-4o mini | openai | $4,932.00 | $0.4932 | $59,184.00 | 128,000 |
| Claude Haiku 4.5 | anthropic | $39,504.00 | $3.95 | $474,048.00 | 1,000,000 |
At 20 queries per user/day with Llama 3.2 3B Instruct, the estimated cost is $0.0504 per monthly active user (MAU). Total monthly bill for 10,000 Active Users is $504.00.
Assuming a 30% cache hit rate on repeated system and context tokens, prompt caching saves $0.00 per month ($0.00/year) on Llama 3.2 3B Instruct.
With a direct COGS of $0.0504 per user/month on Llama 3.2 3B Instruct, charging at least $0.252 per user/month ensures an 80%+ software gross margin.