What the site does
The index currently contains 2,185 distinct models and 8,034 public offers across 53 provider catalogs. They are kept in separate families: 1,444 language models, 48 embeddings, 20 rerankers, 226 image models, 283 video models, 164 audio models. A model page answers three practical questions: which providers sell the model, what they charge for that model type, and what context or capability they report.
The 2,216 active catalog entries include 31 generic provider routes. Routes are listed as provider plumbing but excluded from model totals and price comparisons.
The provider and model pages are generated from a dated snapshot. The comparison surface is intentionally read-only: it does not call a provider API, depend on a database or require an account.
Independence rule. A commercial relationship never changes the order of a price list, the meaning of “lowest”, or the number shown. When an outbound link is commercial, the interface and the disclosure page say so.
How to use it
- Start with the model explorer when you know the capability or price constraint.
- Start with the provider desk when you want to audit a seller’s catalog.
- Open a model sheet to compare every provider offer and raw API id.
- Use the side-by-side board to compare up to three models of the same type; text and media tariffs are never mixed.
What we do not claim
A public price is not a quote, an SLA or a guarantee that a region, account tier or endpoint will be available to you. Prices are point-in-time observations. Always confirm the current terms with the provider before committing a workload.
Read the verification method for the data pipeline and limitations, and report a suspected error through the contact page.