At 200 requests/user/day (totaling 300,000,000 queries/mo), running Data extraction & tagging on DeepInfra — Llama 3.3 70B costs $82,500.00 / month ($1.65 / user / mo).
Total expected API invoice for 50,000 Active Users generating 300,000,000 queries.
Direct inference cost per MAU to model your SaaS pricing tiers and gross margins.
Savings generated by exploiting 20% cache hits on prompt context.
| Model | Provider | Monthly Spend (50,000 Active Users) | Cost / User / Mo | Annual Spend | Context Limit |
|---|---|---|---|---|---|
| DeepInfra — Llama 3.3 70B (Current) | deepinfra | $82,500.00 | $1.65 | $990,000.00 | 128,000 |
| GPT-5.6 Terra | openai | $1,884,000.00 | $37.68 | $22,608,000.00 | 1,050,000 |
| GPT-5.4 Workhorse | openai | $2,355,000.00 | $47.10 | $28,260,000.00 | 256,000 |
| o3-mini | openai | $924,000.00 | $18.48 | $11,088,000.00 | 200,000 |
| o4-mini | openai | $891,000.00 | $17.82 | $10,692,000.00 | 256,000 |
| o1-mini | openai | $924,000.00 | $18.48 | $11,088,000.00 | 128,000 |
At 200 queries per user/day with DeepInfra — Llama 3.3 70B, the estimated cost is $1.65 per monthly active user (MAU). Total monthly bill for 50,000 Active Users is $82,500.00.
Assuming a 20% cache hit rate on repeated system and context tokens, prompt caching saves $0.00 per month ($0.00/year) on DeepInfra — Llama 3.3 70B.
With a direct COGS of $1.65 per user/month on DeepInfra — Llama 3.3 70B, charging at least $8.25 per user/month ensures an 80%+ software gross margin.