About AI Cost Estimator

Estimate what large language model API calls and cloud GPU time will cost, using providers' published price lists, in US dollars or pounds sterling.

Who it is for

Developers, founders and engineering leaders who need a defensible number for what an AI feature will cost: before launch, when choosing between models, or when deciding whether to rent GPUs and run a model themselves.

What it covers

• API prices for 13 large language models from 4 providers, including cached input, batch and long-prompt price tiers where the provider publishes them.
• 32 GPU rental offers across 6 GPU types, with on-demand, spot and reserved prices where published.
• Practical guides to estimating costs, sizing cache and batch savings, and working out the break-even point for self-hosting.

We do not sell API access, GPUs or consulting, and no provider pays to be listed or ranked. The comparison table is always sorted by your own figures.

Ads and partner links

The site is paid for by advertising and, on some model and GPU pages, by partner links. Every ad carries the word "Advertisement" and sits outside the calculator. Every partner link is marked "Partner link" and says that the provider may pay us if you sign up through it. Partner links are listed in the same order as the cost table on that page. Neither ads nor partner links change any price, ranking or result.

How the numbers work

The calculator runs entirely in your browser. It combines the figures you enter with published data. Every data point links to its source and shows the date we last checked it: see the methodology.

If we cannot verify a figure, we leave it out rather than guess.

Who runs this site

This site is run by Muneeb Tariq.

Get in touch

Spotted an error or out-of-date figure? Contact us.