Incorrect Tiered Pricing Calculation (Incremental vs Threshold) #241
Description
Both Python and TypeScript implementations are incorrectly calculating tiered pricing using an incremental model instead of the threshold (cliff) model that is the industry standard used by major providers.
When a request crosses a tier boundary, ALL tokens in that request are charged at the higher tier rate, not just the tokens above the threshold.
See for example Anthropic and Google's docs:
https://cloud.google.com/vertex-ai/generative-ai/pricing#:~:text=Grounding%20with%20your%20data,equivalent%20to%20%240.134%2Fimage%20generated.