At 6 requests/user/day (totaling 9,000 queries/mo), running Customer support chatbot on Mistral Small 4 costs $3.638 / month ($0.0728 / user / mo).
Customer support chatbot for a 50 active users (private beta) audience means 300 requests per day on average, each carrying 3,500 input and 350 output tokens at a 70% cacheable share — so the effective per-request cost on Mistral Small 4 is $0.000404, of which $2.977/mo comes back from prompt caching. Per active user that's $0.0728/mo — the number to hold against your pricing tier and gross margin. Charging at least $0.3638/user/mo keeps gross margin at or above 80%.
Quick answer
For 50 Active Users, Customer support chatbot on Mistral Small 4 is modeled at 9,000 requests per month, costing $3.638 per month or $0.0728 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 50 Active Users generating 9,000 queries.
Direct inference cost per MAU to model your SaaS pricing tiers and gross margins.
Savings generated by exploiting 70% cache hits on prompt context.
| Model | Provider | Monthly Spend (50 Active Users) | Cost / User / Mo | Annual Spend | Context Limit |
|---|---|---|---|---|---|
| Mistral Small 4 (Current) | mistral | $3.638 | $0.0728 | $43.659 | 128,000 |
| GPT-5.6 Luna | openai | $6.111 | $0.1222 | $73.332 | 1,050,000 |
| GPT-5.4 mini | openai | $22.916 | $0.4583 | $274.995 | 256,000 |
| GPT-5.4 nano | openai | $6.269 | $0.1254 | $75.222 | 128,000 |
| GPT-4o mini | openai | $4.961 | $0.0992 | $59.535 | 128,000 |
| Claude Haiku 4.5 | anthropic | $27.405 | $0.5481 | $328.86 | 1,000,000 |
At 6 queries per user/day with Mistral Small 4, the estimated cost is $0.0728 per monthly active user (MAU). Total monthly bill for 50 Active Users is $3.638.
Assuming a 70% cache hit rate on repeated system and context tokens, prompt caching saves $2.977 per month ($35.721/year) on Mistral Small 4.
With a direct COGS of $0.0728 per user/month on Mistral Small 4, charging at least $0.3638 per user/month ensures an 80%+ software gross margin.