Model Price History
A lightweight time series of model pricing, context windows, caching economics, and relevant capability metadata.
Model prices change frequently. Those changes matter because they alter the economics of workflows already in production.
A price history should never overwrite old prices. Each observation should be dated and tied to the source.
Fields
The first version can be a structured file with:
- provider;
- model;
- date observed;
- input token price;
- output token price;
- cached input price;
- context window;
- latency notes;
- relevant capability notes;
- source URL.
This dataset is simple, but it creates historical context for future claims about declining intelligence costs.