An LLM API cost calculator turns a model’s per-token price into the actual dollar cost of a request, a batch of requests, or a whole month of traffic. Large language model providers bill by the token, and they charge a different rate for the tokens you send in than for the tokens the model writes back. The calculator below takes your token counts and the prices you enter, then works out the cost per call and how it scales, entirely in your browser.
What LLM API Pricing Is
LLM providers charge for usage by the token, where a token is a chunk of text roughly three quarters of a word on average. Every API call has two billed parts: the input, which is the prompt and any context you send, and the output, which is the text the model generates in reply. Each part has its own price, quoted per one million tokens. The total cost of a call is the input tokens at the input rate plus the output tokens at the output rate. Because the two rates differ, you cannot estimate cost from a single number; you have to split the call into what went in and what came out.
How to Use It
- Enter the input tokens per request. This is your prompt plus any system message or retrieved context.
- Enter the output tokens per request. This is the length of the reply you expect the model to write.
- Set how many requests are in your run, whether that is one call or a million.
- Type the input and output prices per one million tokens from your provider’s current pricing page.
- Read the total cost and the breakdown. The per-1,000 and per-1,000,000 request rows show how the cost scales with volume.
The Cost Formula
Worked through the defaults above: 1,000 input tokens at $3 per million is $0.003, and 1,000 output tokens at $15 per million is $0.015. The call costs $0.018. One request looks trivial, but the breakdown shows the same call run a million times would cost $18,000, which is why estimating before you scale matters.
Why Output Tokens Usually Cost More
Reading your prompt and writing a reply are not the same amount of work for the model. Input tokens are processed in parallel in a single pass, while output tokens are generated one at a time, each one depending on every token before it. That sequential generation uses more compute per token, so providers price output above input, often by three to five times. The practical takeaway is that long replies cost more than long prompts of the same length, so the length of the answer you ask for is a direct cost lever.
How Cost Scales With Volume
| Requests | Input cost | Output cost | Total |
|---|---|---|---|
| 1 | $0.003 | $0.015 | $0.018 |
| 1,000 | $3.00 | $15.00 | $18.00 |
| 100,000 | $300.00 | $1,500.00 | $1,800.00 |
| 1,000,000 | $3,000.00 | $15,000.00 | $18,000.00 |
Cost is linear in the number of requests, so a small saving on a single call compounds across a high-traffic application. The table shows why teams that run at scale care about token counts that look negligible per call.
How to Reduce Cost
The cheapest token is the one you do not send. Trim system prompts and retrieved context to what the model actually needs, and cap output length when a short answer will do. If you send the same large context repeatedly, prompt caching lets a provider reuse it at a reduced rate instead of charging full price each time. For tasks that do not need a frontier model, a smaller and cheaper model often produces the same result for a fraction of the price. Measure each option with your real token counts in the calculator before you commit.
Prices change often, and every provider lists its own input and output rates. Always read the current cost per million tokens from the provider’s pricing page and enter those numbers here, rather than relying on a figure you saw months ago. To estimate token counts before you have a request to measure, convert your text with our words to tokens tool.
Last Thoughts on LLM API Cost
LLM cost is not a mystery once you split a call into input and output and price each part per million tokens. The formula is fixed; only the prices and token counts change, which is why this calculator lets you edit both and names no models that would go stale. Estimate before you build, then watch the per-million-requests row to see what your design choices cost at scale.
Run your real prompt and expected reply through the calculator above, then compare a long reply against a short one to see the output lever in dollars. For related estimates, convert text length with our word counter, size local hardware with the VRAM for local LLM tool, and browse the rest of our free online tools.
Key Takeaways:
- LLM APIs bill by the token, with separate prices for input and output, both quoted per one million tokens.
- Cost = input_tokens / 1M x input_price + output_tokens / 1M x output_price, times the number of requests.
- Output tokens usually cost more than input tokens because the model generates them one at a time.
- Cost scales linearly with volume, so small per-call savings compound across high traffic.
- Cut cost with shorter prompts, capped output length, prompt caching, and smaller models where they suffice.
- Enter prices from the provider’s current pricing page, since rates change and differ by provider.
Frequently Asked Questions (FAQs)
How is LLM API cost calculated?
Take the input tokens, divide by one million, and multiply by the input price per million. Do the same with the output tokens and the output price. Add the two figures for the cost of one request, then multiply by the number of requests for the cost of a run. The calculator on this page does all of this once you enter the five values.
Why are input and output tokens priced differently?
Input tokens are read in a single parallel pass, while output tokens are generated one after another, each depending on the ones before it. That sequential work uses more compute per token, so providers charge more for output, often several times the input rate. This is why a long reply costs more than a long prompt of the same length.
What is a token?
A token is a chunk of text the model processes as a unit, averaging about three quarters of a word in English. Short common words are one token, while longer or rarer words split into several. Both your prompt and the model’s reply are measured in tokens, and you are billed for both.
Does this calculator use live model prices?
No, and that is deliberate. You type in the input and output prices yourself, so the tool never relies on a figure that has gone out of date and names no specific model. Copy the current cost per million tokens from your provider’s pricing page into the two price fields for an accurate result.
How can I lower my LLM API bill?
Send fewer tokens by trimming prompts and context, and limit output length when a short answer is enough. Use prompt caching to reuse repeated context at a lower rate, and pick a smaller model for tasks that do not need a frontier one. Measure each change with your real token counts before you roll it out.
Is anything I enter sent to a server?
No. The calculation runs entirely in your browser using the numbers you type. Nothing is uploaded, logged, or stored, so you can estimate costs for private prompts and internal pricing without anything leaving your device.
