Blended Price
A single headline figure that averages input and output rates so you can rank models at a glance.
A blended price collapses a model’s separate input and output rates into one number so models can be sorted and scanned quickly. On this explorer, the blended figure is the simple average of the input and output rate per 1M tokens — a deliberately transparent definition, not a weighted or usage-based estimate.
Blended prices are useful for a first-pass ranking, but they hide the mix. A model with cheap input and pricey output has the same blend as one with the reverse, yet the two cost very different amounts on a real workload — the first wins on reading-heavy tasks, the second on writing-heavy ones. Any blended number that assumes a fixed input:output ratio bakes in an assumption that may not match your traffic, which is why we keep ours to a plain average and label it.
For an accurate figure, use the token-cost calculator on the hub: enter your real input and output token counts, cache rate, and monthly volume, and it ranks every model on your actual mix rather than an averaged proxy. Treat the blended price as a way to shortlist candidates, then confirm with the calculator before deciding. Honest ranking beats a false-precision single number.
FAQ
How is the blended price calculated here?
As a simple average of the input and output rate per 1M tokens. It is a ranking aid, not a workload estimate, and it assumes no particular input/output mix.
Should I choose a model on blended price alone?
No. Blended price hides the input/output split. Use it to shortlist, then run your real token mix through the hub calculator to see the true cost.