At 15 requests/user/day (totaling 1,125,000 queries/mo), running Translation on GPT-4o mini costs $4,492.969 / month ($1.797 / user / mo).
Translation for a 2,500 active users (series a product) audience means 37,500 requests per day on average, each carrying 5,000 input and 5,500 output tokens at a 15% cacheable share — so the effective per-request cost on GPT-4o mini is $0.003994, of which $63.281/mo comes back from prompt caching. Per active user that's $1.797/mo — the number to hold against your pricing tier and gross margin. Charging at least $8.986/user/mo keeps gross margin at or above 80%.
Quick answer
For 2,500 Active Users, Translation on GPT-4o mini is modeled at 1,125,000 requests per month, costing $4,492.969 per month or $1.797 per active user.
Method & trust
This projection combines the selected active-user tier, requests per user per day, input/output tokens per request, and the modeled cache share. It is an operating estimate, not a provider invoice; production traffic and retries should be measured against it.
Total expected API invoice for 2,500 Active Users generating 1,125,000 queries.
Direct inference cost per MAU to model your SaaS pricing tiers and gross margins.
Savings generated by exploiting 15% cache hits on prompt context.
| Model | Provider | Monthly Spend (2,500 Active Users) | Cost / User / Mo | Annual Spend | Context Limit |
|---|---|---|---|---|---|
| GPT-4o mini (Current) | openai | $4,492.969 | $1.797 | $53,915.625 | 128,000 |
| GPT-5.6 Luna | openai | $8,398.125 | $3.359 | $100,777.50 | 1,050,000 |
| GPT-5.4 mini | openai | $31,492.969 | $12.597 | $377,915.625 | 256,000 |
| GPT-5.4 nano | openai | $8,707.50 | $3.483 | $104,490.00 | 128,000 |
| Claude Haiku 4.5 | anthropic | $35,803.125 | $14.321 | $429,637.50 | 1,000,000 |
| Claude 3.5 Haiku | anthropic | $28,642.50 | $11.457 | $343,710.00 | 200,000 |
At 15 queries per user/day with GPT-4o mini, the estimated cost is $1.797 per monthly active user (MAU). Total monthly bill for 2,500 Active Users is $4,492.969.
Assuming a 15% cache hit rate on repeated system and context tokens, prompt caching saves $63.281 per month ($759.375/year) on GPT-4o mini.
With a direct COGS of $1.797 per user/month on GPT-4o mini, charging at least $8.986 per user/month ensures an 80%+ software gross margin.