About and contact

Anyone can read today's model prices off a provider's own page. Almost nobody kept yesterday's. LLMPrice.io exists to keep them, and to say what they mean for a bill somebody actually has to pay.

Two things are published here. The first is an archive: 175 captures of 2,988 models between 2024-01-01 and 2026-08-12, each one a record of what every provider was charging on the day we looked. The second is four monthly price indices built from it, which price the same twelve models on four different workload shapes, so a reader can see whether the kind of work they run has actually got cheaper rather than whether the headline model has.

Everything else on the site is a tool built on top of those two: an estimator that turns plain answers into a monthly figure, a bill auditor that re-prices a usage export across every model we track, and a rate card.

How the figures are produced

Rates are captured nightly from the LiteLLM public pricing dataset, a continuously maintained community record of provider list prices, and committed to our own repository. We do not retype figures by hand and we do not scrape provider pages. That means the archive inherits that dataset's coverage and its mistakes; what we add is time, because the dataset publishes what is true now and replaces it, while we keep what it said on each date.

Every published number here is either a rate we observed or arithmetic on one. Where we cannot compute something honestly, we say so instead of estimating it: models whose provider publishes no image tokenisation rule are excluded from image pricing rather than given a plausible figure, and the completeness of the archive, including the days it does not hold, is stated in full on the archive page.

What we do not publish

No quality scores and no benchmarks. We measure price, and nothing else. We publish no intelligence ranking, scrape none, and republish no third party's composite. Where the site needs an ordering of models it uses the provider's own output price as a stated proxy, and says that is what it is.

No paid placement and no affiliate arrangements. Nothing in any ranking is sponsored, and no link on this site earns us a commission. A cheaper model ranks above an expensive one because it is cheaper.

Method in the open. How the indices are constructed, how the basket is maintained, how substitutions are chain-linked and how corrections are handled is written out on the methodology page. Published issues are frozen on release and annotated rather than quietly edited, so a citation of a dated issue still resolves to the figures that were published.

Corrections

Found a pricing error? Email a link to the provider's official page and we will check it against the archive and correct it. If a figure here disagrees with the provider, the provider is right and we want to know. The site is small enough that a real person reads every message.

Who publishes this

LLMPrice.io is published by Sugarpine Property Management LLC. It is a small, independently run project rather than a company product, which is why the contact below is a person rather than a ticket queue.

Contact: questions, corrections, or partnership inquiries: aj@sugarpinepm.com
Sugarpine Property Management LLC · 637 S Broadway St STE B #218, Boulder, CO 80305 · (303) 335-9399