Cargando...
Cargando...
We use our own cookies to make the site work, Google cookies to show you advertising, and analytics cookies to find out which pages are read and which are not, in aggregate. If you decline, ads are still shown, but not personalised. You can accept everything or keep only the necessary ones.
Four tools that do arithmetic on the prices our articles talk about, so you can check our numbers and work out your own. No account, no email, no lead capture. Everything runs in your browser.
Turns tokens into real money. Enter how many input and output tokens your task uses and get the cost of one run, the monthly cost, and how many runs fit your budget.
Data for 13 models, each with its own verification date.
Open tool →Compares up to five models side by side on seven objective criteria: price, context window, max output, caching, batch discount, long-context pricing and release date.
Prices and specifications published by each provider.
Open tool →Works out the return on AI spend from your real hourly cost and the tasks you automate. Adjust the hours per task and the share AI removes, and get the saving and the break-even point.
Your own numbers. The tool does not invent the saving percentage.
Open tool →Works out what image and video generation costs, counting outputs per generation, adjustment passes, and the fact that you are billed for the whole clip even if you use four seconds of it.
Per-image and per-second prices published by each provider.
Open tool →We write about AI pricing a lot, and we kept finding the same problem: readers could not tell whether a price we quoted applied to their situation. Per-million-token prices are useless on their own. They only become meaningful once you know how many tokens your own task uses.
Price data last verified on 2026-09-28. If a provider changes their pricing after that date, the calculators will be behind until we re-check.
Quick example
1,050,000-token context · max 128,000 output · knowledge cutoff 2026-04-30
1,200
0
400
Cost of one run
€0.028
= $0.03 · 1,600 tokens in total
Monthly cost (1,000 runs)
€28.0431
$32.00
With $100.00 per month
3,125
runs per month
Price source
OpenAI — Pricing (API)Verified: 2026-09-28
Exchange rate applied: 1 $ = 0.8763 € (Banco Central Europeo, referencia del 23/09/2026, 2026-09-23)
Choose the models to compare
| Model | Score | Input | Output | Cached input | Context | Max output | Cutoff |
|---|---|---|---|---|---|---|---|
| GPT-6 LunaOpenAI | 8.4/10 | $0.10 | $0.50 | $0.01 | 1.050.000 | 128.000 | 2026-04-30 |
| GPT-6 SolOpenAI | 8.1/10 | $2 | $10 | $0.20 | 1.050.000 | 128.000 | 2026-04-30 |
| Claude Fable 5.1Anthropic | 7.1/10 | $10 | $50 | $0.25 | 1.000.000 | 128.000 | 2026-06-30 |
| Claude Mythos 5.1Anthropic | 7.1/10 | $10 | $50 | $0.25 | 1.000.000 | 128.000 | 2026-06-30 |
| GPT-6 AstraOpenAI | 7.0/10 | $10 | $50 | $1 | 1.050.000 | 128.000 | 2026-04-30 |
How much each criterion matters to you
Set a criterion to 0 and the score ignores it. If price is all you care about, set only that to 1 and the rest to 0: you get the cheapest ranking, with no invented criterion sneaking into the average.
Scores the balance between capability and cost, not raw power.
How much fits in one prompt: a contract, a repository, a book.
Matters if you generate long documents or whole reports in one go.
Price per million tokens. Lower is better here.
How much the price drops when you reuse the same context repeatedly.
What you save if you can wait instead of needing the answer right now.
Matters if you work with regulations, prices or products that change every few months.
How the score is calculated
It is the weighted mean of the criteria with a non-zero weight, each normalised to 0-10 across every model in the price table. It is not a lab benchmark: it is a transparent average of figures you can check in each provider's documentation. Switch the sort to "Price" if you want to minimise spend rather than maximise capability.
Prices per million tokens, verified on 2026-09-28 against each provider's official documentation. Exchange rate 1 $ = 0.8763 € (2026-09-23).
Profiles are starting points. Replace the hourly cost with your own — that is the only input that makes the result yours.
You are using the reference figure for "Technical profile" (€32/hour).
Adjust the tasks to your reality
Weekly hours you spend today, and the share of that task AI can realistically remove. Be honest about the second figure: if you put 80% where reality is 20%, the result is worthless.
Net saving over the period
€-506.40
353 hours recovered over 12 months
ROI on the tool spend
-16 %
On these figures the tool spend does not pay off. Lower your tool cost or revisit the hours.
Monthly break-even
€1
Above this tools spend the operation turns profitable.
Weekly hours freed
6.8 h
Out of 19.0 h/week you report on automatable tasks.
Real cost of your hour
€34.22
It used to cost €32.00 per task hour.
What this calculator does not do
How many images you actually get per prompt.
How many versions you discard per image you keep.
Total monthly cost
€2.80
$3.20 per month
Details that change the number
Price sources
No. They are static pages and all the maths happens in your browser. There is no account, no server call and no form.
Because we check prices on a set date and record it on every row. If the provider has changed since, our number is the older one. The verification date is shown next to each price so you can see how fresh it is.
Because most readers here pay in euros. We use the European Central Bank reference rate and we show the exact rate and date rather than hiding it in the calculation.
Yes. That is what they are for. But replace the reference figures with your own before you present anything: the ROI calculator is only as good as the hourly cost you enter.