LLM API Pricing Calculator
Calculate and compare costs for different LLM providers and models. Estimate your AI API expenses based on token usage.
What is the LLM API Pricing Calculator?
The LLM API Pricing Calculator is a free tool that estimates and compares the cost of an API call across dozens of large language models. Enter how many input and output tokens you expect, and it produces a side-by-side cost table.
It covers models from OpenAI, Anthropic, Google, X AI, DeepSeek, Meta, Mistral, and Moonshot AI, each with its published per-million-token input and output rates. The table sorts from cheapest to most expensive and tags the lowest-cost option so comparison is immediate.
The calculator runs entirely in your browser with no signup. Prices are based on standard API rates as of late December 2025 and may vary by region, cache usage, or enterprise agreements.
How to use the LLM API Pricing Calculator
- 1Enter the number of input tokens you expect to send in a request, such as your prompt and context.
- 2Enter the number of output tokens you expect the model to generate in its completion.
- 3Optionally use the provider filter to narrow the table to a single vendor like OpenAI or Anthropic.
- 4Read the Cost Comparison table, which lists each model's input, cached, and output rates alongside the total estimated cost.
- 5Look for the row tagged Cheapest, which is the lowest-cost model for the token counts you entered.
What you can use it for
- Comparing the per-call cost of GPT-5, Claude, Gemini, and other models before choosing one for a feature.
- Estimating the monthly bill for an AI feature by multiplying a single-call cost by expected request volume.
- Deciding whether a smaller or mini model is cheap enough to handle high-volume, low-complexity tasks.
- Checking how much a long-context prompt costs when input tokens are large relative to output.
- Building a budget proposal that needs concrete dollar figures for several candidate models.
Key features
- Compares dozens of models across eight providers in one sortable table.
- Calculates total cost from separate input and output token counts using per-million-token rates.
- Automatically sorts results cheapest first and labels the lowest-cost model.
- Shows context window sizes and cached input rates where the provider publishes them.
- Filter the comparison to a single provider when you only want to weigh one vendor's lineup.
Frequently asked questions
How is LLM API cost calculated?
Cost is the input tokens divided by one million times the model's input rate, plus the output tokens divided by one million times the output rate. This calculator applies that formula to every model so you can compare total estimated costs for the same workload.
What is the difference between input and output tokens in API pricing?
Input tokens are what you send to the model, including your prompt, system message, and context. Output tokens are what the model generates in response. Output tokens are usually priced several times higher than input tokens, so the input-to-output ratio strongly affects total cost.
Which LLM API is the cheapest?
There is no single cheapest model because the answer depends on your input-to-output token ratio. Enter your expected token counts and the calculator sorts every model from lowest to highest cost and tags the cheapest one for that specific workload.
How many tokens is one word or one page of text?
As a rough guide, one million tokens is roughly 750,000 words, so a token averages a bit under one word. Actual counts vary by language and content, so treat token estimates as approximate when budgeting.
Are the prices in the LLM API pricing calculator up to date?
The prices reflect standard published API rates as of late December 2025. Providers change pricing over time, and actual costs can differ with region, prompt caching, or enterprise agreements, so confirm current rates on the provider's official pricing page before committing.
What is cached input pricing and does the calculator use it?
Cached input pricing is a reduced rate some providers charge when a prompt prefix is reused from cache. The table displays the cached rate where a provider offers one, but the total estimated cost is calculated using the standard input rate, not the cached rate.