At 6 requests/user/day (totaling 90,000 queries/mo), running Customer support chatbot on GPT-5.3 Codex costs $644.963 / month ($1.29 / user / mo).
Customer support chatbot for a 500 active users (early growth) audience means 3,000 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 GPT-5.3 Codex is $0.007166, of which $347.287/mo comes back from prompt caching. Per active user that's $1.29/mo — the number to hold against your pricing tier and gross margin. Charging at least $6.45/user/mo keeps gross margin at or above 80%.
Quick answer
For 500 Active Users, Customer support chatbot on GPT-5.3 Codex is modeled at 90,000 requests per month, costing $644.963 per month or $1.29 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 500 Active Users generating 90,000 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 (500 Active Users) | Cost / User / Mo | Annual Spend | Context Limit |
|---|---|---|---|---|---|
| GPT-5.3 Codex (Current) | openai | $644.963 | $1.29 | $7,739.55 | 256,000 |
| DeepSeek Coder V2.5 | deepseek | $25.137 | $0.0503 | $301.644 | 128,000 |
| Codestral 2501 | mistral | $63.315 | $0.1266 | $759.78 | 256,000 |
| Grok Build 0.1 | xai | $201.60 | $0.4032 | $2,419.20 | 256,000 |
| Qwen 2.5 Coder 32B | qwen | $42.21 | $0.0844 | $506.52 | 128,000 |
| GLM-5.3 (Z.ai) | zai | $328.23 | $0.6565 | $3,938.76 | 1,048,576 |
At 6 queries per user/day with GPT-5.3 Codex, the estimated cost is $1.29 per monthly active user (MAU). Total monthly bill for 500 Active Users is $644.963.
Assuming a 70% cache hit rate on repeated system and context tokens, prompt caching saves $347.287 per month ($4,167.45/year) on GPT-5.3 Codex.
With a direct COGS of $1.29 per user/month on GPT-5.3 Codex, charging at least $6.45 per user/month ensures an 80%+ software gross margin.