The main takeaway is simple, Frontier model costs did not move: the latest blended price is $4.64 per 1M tokens, and the change over the 120-day window is 0% from $4.64. For teams budgeting around flagship models, that means the cost baseline stayed stable across the period.
Budget models are still much cheaper at $0.74 per 1M tokens, so the spread between Frontier and lower-cost usage remains the key planning variable, not short-term volatility.
The AI Cost Index (Frontier basket), blended $/1M tokens over time.
This index is built on a fixed Frontier basket of flagship models with a 3:1 input:output blend, so the flat line is telling you something useful, not just that prices were quiet. It says the blended cost of running these models over time was stable enough that a developer could treat $4.64 per 1M tokens as a steady budgeting anchor during the window.
That matters because AI spend usually gets messy when prices swing, especially for products with predictable token volume. Here, the data shows no drift in the Frontier basket across 120 daily points, which reduces the need for frequent budget resets and makes margin planning easier.
The Budget basket at $0.74 per 1M tokens shows the other side of the equation, cheaper model usage is still available at a much lower blended rate. For developers, the practical implication is that architecture choices, not market movement, are likely to dominate cost control.
Frontier stayed flat at $4.64 per 1M tokens across the window.
Budget is priced at $0.74 per 1M tokens, far below Frontier.
Use the stable Frontier price as a budgeting baseline, not a moving target.
Model selection will matter more than price volatility for cost control.
About this data. Figures are from MyTokenTracker live data (AI Cost Index, open data API), free to use and cite under CC BY 4.0. Snapshot: October 19, 2026.
