What does my AI stack cost per month?
Pick your tools and plans. You see the total per month and per year right away, and where it could be cheaper.
Calculate subscription costs
Many AI tools charge a fixed monthly or yearly subscription. The calculator below adds up the list prices of your plans and shows plans from other vendors that, according to the vendor, include at least two thirds of the selected features.
Calculate API usage
Many AI tools use subscriptions. Others charge by usage or provide prices only on request. The calculator below estimates API costs from your inputs for selected language models.
Using this calculator is free. No account or subscription is required. More on free use
Subscription vs API: what would your usage cost on the API?
Pick a subscription and roughly describe how much you write with the AI. The calculator estimates what the same usage would cost on the API and shows whether the subscription or the API is cheaper.
Catalog price data
AI subscription calculator
Your stack
No tools selected yet.
Filters
Alternatives from other vendors
Shown: plans from other vendors that, according to the vendor, include at least 6 of the 9 selected features, grouped by performance level. Savings are only calculated within the same level. Specialist plans, such as coding-only ones, appear only when your plan has the same focus or you select that feature specifically.
✓ included according to the vendor · − not included according to the vendor · ? not stated by the vendor
Comparison by vendor-stated features and list price. It says nothing about quality, model performance or exact usage limits. Check the vendor’s terms before you switch.
How the selection works: we show plans from our catalog that, according to the vendor’s page, include at least two thirds of the selected features, grouped by performance level (Basic, Standard, Power, Pro/Max) and sorted by match and price or by the order you choose. We derive the level from the vendor’s published usage limits; where the vendor states none, the list price serves as a fallback signal. We do not cover every vendor on the market.
Calculated only in your browser.
As of: check date per plan. Check binding prices with the provider. Prices as stated by the provider, usually plus statutory VAT. The final price depends on your country and is shown by the provider. Regional prices, discounts and usage costs may differ. Yearly prices are shown per month. Terms of use
Model calculation
Subscription vs API
If the plan also includes API credits according to the provider, they are not included here.
What your subscription price would buy on the API
This does not say how many tokens the subscription includes. The provider does not publish such a number.
Subscription usage limit
Token limit: unknown. The provider does not state a token number.
On the API, this usage would have cost more than ten times the subscription price. The provider does not say in numbers whether the subscription allows that much use. Expect that a limit may apply.
Assumptions behind this calculation
Model calculation based on your assumptions and public list prices. As of: see the check dates in the assumptions above. Check binding prices with the provider. Actual costs and limits may differ.
How the comparison works
Cost per message = input tokens × input price (with cached share) + (output tokens + reasoning tokens) × output price. Times messages per day and times 30 days gives the month. We compare this amount with the fixed subscription price. We do not estimate token allowances for subscriptions, because providers do not publish them.
How to read the result
The monthly total converts yearly plans to a monthly figure. Per-seat plans count the number of seats. Free plans count as 0.
What is the calculator for?
A quick overview of your recurring AI costs, and where you can get similar features for less. It compares list prices and vendor-stated features, not quality.
What does a run with several API requests cost?
In the API calculator: requests per month = API requests per run × runs per month. A fixed cap per run keeps API costs predictable.
How to read the inputs
A request is one call to a model. Input tokens are the text you send, and output tokens are its answer. Use averages from a typical prompt and response.
What is the calculator for?
Use it, for example, to plan budgets for support bots, document analysis, coding assistants, or similar. It compares published unit prices, not a model's quality or suitability.
How the cost estimate works
Key answers in plain language, to help you get a realistic first budget figure.
Where do the API prices come from?
They come from OpenRouter's public price list. You therefore see what a model costs through OpenRouter. That can differ from the price charged directly by the model maker. The source and data timestamp appear at the bottom of the calculator.
How fresh is the data?
The calculator tells you when it loaded the model data. If an update fails, it uses the last confirmed data and says so clearly.
How is the monthly price calculated?
Simply: your monthly requests × the average length of a question and answer × the model price. More or longer conversations raise the amount. Extra features are included only when you enable them and a price is available.
What do the extra inputs mean?
Input is your text and output is the model's reply. You can optionally include reused text from a cache, a model's internal reasoning work, images, audio or web searches. Use those fields only when they fit your project.
What happens with reused text?
Some providers charge less when text can be reused from a cache. If a published special price exists, the calculator uses it. Otherwise it cautiously uses the normal input price rather than assuming a discount.
Does the calculator include special prices?
Yes, but only when OpenRouter publishes them for that exact access route, for example based on text length or time. Unlisted discounts and special terms are never guessed.
Why does it check technical limits?
So a low estimate is not shown for something that cannot work. The calculator checks whether the question and reply fit the model, and whether extras such as images, audio or web search are supported.
What is not included in the price?
The result is a budgeting aid, not an invoice. Taxes, your individual contract, account limits, minimum commitments and unpublished discounts are not included. If something has no price, the calculator does not show a deceptively exact amount.
Is the cheapest model automatically the best?
No. The calculator compares costs, not quality. Also check whether a model really suits your language, tasks, preferred reply length and features.
Compare models directly
Compare prices, limits and sources for common models side by side.
Frequently asked questions
Where do the prices come from?
From our catalog: list prices from the vendors' official information, excluding tax. Regional prices, discounts and usage costs may differ.
Are my entries stored?
No. The calculator runs only in your browser. The tools you pick are neither stored nor sent.
How are the cheaper suggestions found?
The calculator looks for cheaper plans of the same tool and comparable tools from the same category in our catalog. Whether they suit you is your call.
Can I embed the calculator on my website?
Yes, with the Embed button below the calculator. The embedded version also calculates only in the browser and loads no tracking services.