AI, by the numbers
AI API Cost Calculator
Price a workload across today's models — per request, per day, per month.
Per request
$0.013
Per day
$13.50
Per month
$405
Same workload, every model
Sorted cheapest first — monthly cost at your volume.
| Model | / req | / day | / month |
|---|---|---|---|
| Gemini 3.1 Flash-LiteGoogle | $0.0013 | $1.25 | $37.50 |
| GPT-5.4 miniOpenAI | $0.0038 | $3.75 | $113 |
| Haiku 4.5Anthropic | $0.0045 | $4.50 | $135 |
| GPT-5.6 LunaOpenAI | $0.0050 | $5.00 | $150 |
| Gemini 3.5 FlashGoogle | $0.0075 | $7.50 | $225 |
| Gemini 3.1 ProGoogle | $0.010 | $10.00 | $300 |
| GPT-5.6 TerraOpenAI | $0.013 | $12.50 | $375 |
| Sonnet 5Anthropic | $0.013 | $13.50 | $405 |
| Opus 4.8Anthropic | $0.022 | $22.50 | $675 |
| GPT-5.6 SolOpenAI | $0.025 | $25.00 | $750 |
| Fable 5Anthropic | $0.045 | $45.00 | $1,350 |
Prices verified July 12, 2026, USD per 1M tokens. Estimates only — excludes caching, batch discounts, and image/audio tokens.
How this works
Cost per request is (input tokens ÷ 1M × input price) + (output tokens ÷ 1M × output price), then multiplied by your daily volume and by 30 for a month. The table runs the same workload through every model so you can see the spread — which is often 10× or more between a flagship and a small model.
FAQ
Why is output so much more expensive? Generating tokens is more compute-intensive than reading them, so output is typically 4–6× the input price. Workloads that emit long responses are dominated by output cost.
What's not included? Prompt caching and batch APIs can cut costs substantially; this is the list-price estimate before those.