Method
Methodology and sources
One index, one version
Every score, price, cost per task, time per task and token count comes from Artificial Analysis, Intelligence Index v5.0.1, retrieved November 12, 2026. Values are taken at each model's highest available reasoning effort, named in every table. Input modalities and context windows come from the technical-specification section of each model's own Artificial Analysis page.
Retrieved, not transcribed
The 18-model table is generated by a script from the source pages, not typed by hand. Each cell keeps a verbatim excerpt of the page it came from, so a claim on this site can be traced to a specific sentence on a specific page. A second source — the OpenRouter catalogue — is used to cross-check that a model exists and to flag price differences; it is never averaged with the first, because two sources recording different things do not produce a third fact.
What the labels mean
- N/A (AA: Unknown) — the source publishes no value in that cell. It is not zero, and it is not "unreleased".
- not applicable (AA prints
-) — the concept does not exist for that model (for example, no per-token price when there is no token billing). - not computable (…) — the cell is derived from two others and one is missing; the bracket names which.
- not printed on its AA page — the page lists prices but withholds that figure.
What we exclude
Models that appear in one aggregator but cannot be confirmed in a second source, and models marked unreleased or preview-only. A missing cost is not grounds for exclusion: the model stays, with the gap labelled.
Update policy
Pages are rebuilt weekly. If a model's index moves by 2 points or more, or a new model enters the top 50, the table and the updated date change together. If nothing changes, the date still moves — a timestamp that never moves is how comparison sites die.