Author
I'm a computer engineering student in Madrid. I work at a technology consultancy and I'm building NeuralTech, a small AI and technology consulting practice of my own in Madrid. I pay for LLM API usage out of my own pocket on those projects, and that bill is the reason this site exists.
Why this site
Picking a model used to mean opening six pricing pages and hoping none of them had changed since the last time I looked. Providers raise prices, quietly end promotional rates, rename models and retire endpoints, and most of the comparison tables you find in a search are already months out of date — which is worse than having no table at all, because a stale number still looks like a fact.
So I built the one I wanted. Every price here was read off the provider's own pricing page, and every row carries the date it was checked and a link to the page it came from, so you can verify any figure yourself in one click.
How I work
- No estimated prices. If a figure can't be confirmed against an official source, the row says "no data" and links to the provider instead of showing a number; where the maker publishes no per-token price at all, the row says "no published price", dated the day that was checked. Right now five current rows say "no published price" and zero say "no data", rather than carry a guess.
- Every number is dated. A price without a verification date is an opinion. Each row shows when it was last checked and against which page.
- Changes are logged, not overwritten. When a provider moves a price, the old figure stays on the record so you can see what changed and when.
- Promotional pricing is labelled with its end date. A promo rate is a real price today and a budgeting trap in three months, so both the current and the list price are shown.
- I use AI tooling to build and maintain this site, and I review what it produces. Nothing about pricing gets published because a model asserted it — it gets published because it was checked against the provider's page.
Corrections
If a number here is wrong, I want to know. Email me at info@neuraltech.pro with the model and the source you're looking at, and I'll check it and correct it. Corrections are the whole point of publishing the sources.