LLM API cost calculator
Enter your model's price per million input and output tokens, your average tokens per request, and your daily request volume; get cost per request, per day, and per month — with an optional prompt-caching discount.
1,800.00 per month at 5,000 requests/day.
- Cost / request
- 0.01
- Cost / day
- 60.00
- Cost / month
- 1,800.00
Output tokens usually cost several times more than input. Prices change often — always paste current provider prices.
How it works
You supply the prices yourself: cost per one million input tokens and per one million output tokens, taken from your provider's current pricing page. The calculator deliberately hardcodes nothing — model prices change too often for baked-in numbers to stay honest, and a calculator quoting stale prices is worse than no calculator.
From your average input and output tokens per request, it computes cost per request as (input tokens times input price plus output tokens times output price) divided by 1,000,000. Multiplying by requests per day and roughly thirty days gives daily and monthly figures. Output tokens usually cost several times more per token than input, so the input/output split matters more than the total.
The optional caching section models prompt caching: you set what percentage of input tokens are served from cache and the price multiplier your provider applies to cached tokens. Cached input is billed at the discounted rate, the remainder at full price — a realistic model for workloads with a large shared system prompt or repeated context.
Assumptions and limitations
- The math is only as good as your token estimates. Real traffic has a distribution, not an average, and a heavy tail of long requests can dominate the bill — check your provider's usage dashboard against these projections early.
- Caching behavior differs by provider: minimum cacheable lengths, time-to-live windows, and cache-write surcharges all affect the real discount. This calculator models the steady-state read discount only.
- It covers token costs, not the full bill — rate limit tiers, batch pricing, image or audio tokens, and fine-tuned model premiums are out of scope.
- Prices you enter are a snapshot. Re-check them whenever you revisit a projection; providers reprice without much notice.
Frequently asked questions
How do I calculate the cost of an LLM API call?
Multiply input tokens by the input price per million, multiply output tokens by the output price per million, divide each by one million, and add them. The two rates are almost always different — output tokens typically cost several times more — so estimating the split matters as much as estimating the total.
Why doesn't this calculator include current model prices?
Because they change frequently enough that any hardcoded list would mislead someone within months. You paste prices straight from your provider's pricing page, which keeps the calculation trustworthy regardless of when you use the tool or which model you are evaluating.
How much does prompt caching reduce LLM API costs?
It depends on two numbers you control in this calculator: the share of input tokens served from cache and the cached-token price multiplier your provider charges. Workloads with a long, stable system prompt and short user turns see the biggest reduction; workloads where every request is unique see almost none.
How do I find out how many tokens my requests use?
The most reliable source is the usage metadata your provider returns with each API response, which reports exact input and output token counts. Log those for a representative sample of real traffic and use the averages here. For rough planning before you have traffic, a heuristic like four characters per token gets you in the ballpark.
Are the prices and volumes I enter here sent anywhere?
No. All arithmetic runs client-side in your browser. Your pricing, token counts, and request volumes are never transmitted, logged, or stored.
More free tools like this, by email
I publish new calculators and explainers regularly. Get them when they land — no spam, unsubscribe anytime.