What it is
One table of list prices for language-model APIs: the vendor's own API, the cloud platforms that host the model, and the marketplaces that resell it. Prices are per 1 million tokens in the channel's currency; input, output, cached and batch rates are recorded wherever the channel publishes them.
What makes it different
- Every number has a source. Each price links to the page or price list it was read from, with the fetch time and a hash of what was seen.
- Regions, not just one price. Cloud platforms charge different prices by region. The index records each region separately instead of showing one “global” figure.
- History is kept. Records are added, never edited. When a price changes, the old record stays.
- Status is explicit. A price that an independent second reader has re-read from the snapshot, blind to the recorded value, is marked verified; everything else is marked unverified. Nothing is hidden behind a badge.
What it is not
Not affiliated with any vendor. Not a reseller and not a marketplace. It does not include contract prices, taxes, or prices from unofficial proxies. It is not purchasing advice: the channel's own page is always authoritative.
Who runs it
AuraMatrix. Something wrong, or a question? Write to hello@auramatrix.ai. The method is public: Methodology.