LLM API Cost Calculator
Estimate what an LLM feature will actually cost to run. Enter your tokens per request and daily volume, and compare monthly spend across models. Every price is editable — match them to your provider's current rates.
Estimated cost by model
| Model | $/1M in | $/1M out | Per request | Per day | Per month | |
|---|---|---|---|---|---|---|
| Gemini 3 FlashCheapest | <$0.01 | $2.50 | $75.00 | |||
| Claude Haiku 4.5 | <$0.01 | $4.50 | $135 | |||
| GPT-5.6 Luna | <$0.01 | $5.00 | $150 | |||
| Gemini 3.5 Flash | <$0.01 | $7.50 | $225 | |||
| Claude Sonnet 5 | <$0.01 | $9.00 | $270 | |||
| Gemini 3.1 Pro | $0.01 | $10.00 | $300 | |||
| GPT-5.6 Terra | $0.01 | $12.50 | $375 | |||
| GPT-5.4 | $0.01 | $12.50 | $375 | |||
| Claude Opus 4.8 | $0.02 | $22.50 | $675 | |||
| GPT-5.6 Sol | $0.03 | $25.00 | $750 | |||
| GPT-5.5 | $0.03 | $25.00 | $750 |
Monthly = per-day × 30. Seed rates are mid-2026 examples ($ per 1M tokens) and change often — always verify against your provider's current pricing page and edit the cells above. Estimates exclude cached-input discounts, batch pricing, and taxes.
How LLM API cost works
Most providers bill per token — roughly ¾ of a word — and charge more for the tokens a model generates (output) than the ones you send it (input). Your real bill is driven by three levers: how much context you send, how much the model writes back, and how many requests you serve. Trimming prompt size and capping output length usually cuts cost faster than switching models.
Go deeper
Build with AI? Get the daily rundown.
Original AI briefings for founders, PMs, and builders — free, every morning.