Models
The top model releases by token volume, each priced as the weighted median across the providers that serve it, weighted by the tokens each served, with each provider capped at 35% where three or more serve it (with one or two the cap cannot hold, and the row says so). USD per 1M tokens.
| # | Release | Input | Output | Cache read | 30-day trend | Cost / task | AA index | Providers | Spread | Weighting |
|---|
Price map: blended price against capability
Each bubble is a release. Blended price is the 50/50 average of input and output; capability is the AA Intelligence Index. Log price scale.