At 10 requests/user/day (totaling 150,000 queries/mo), running RAG / search-augmented answers on Together AI — GPT-OSS-20B costs $75.00 / month ($0.15 / user / mo).
Total expected API invoice for 500 Active Users generating 150,000 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 (500 Active Users) | Cost / User / Mo | Annual Spend | Context Limit |
|---|---|---|---|---|---|
| Together AI — GPT-OSS-20B (Current) | together | $75.00 | $0.15 | $900.00 | 32,768 |
| GPT-5.6 Luna | openai | $222.00 | $0.444 | $2,664.00 | 1,050,000 |
| GPT-5.4 mini | openai | $832.50 | $1.665 | $9,990.00 | 256,000 |
| GPT-5.4 nano | openai | $225.75 | $0.4515 | $2,709.00 | 128,000 |
| GPT-4o mini | openai | $180.00 | $0.36 | $2,160.00 | 128,000 |
| Claude Haiku 4.5 | anthropic | $1,035.00 | $2.07 | $12,420.00 | 1,000,000 |
At 10 queries per user/day with Together AI — GPT-OSS-20B, the estimated cost is $0.15 per monthly active user (MAU). Total monthly bill for 500 Active Users is $75.00.
Assuming a 50% cache hit rate on repeated system and context tokens, prompt caching saves $0.00 per month ($0.00/year) on Together AI — GPT-OSS-20B.
With a direct COGS of $0.15 per user/month on Together AI — GPT-OSS-20B, charging at least $0.75 per user/month ensures an 80%+ software gross margin.