DictionaryToken Economics
Long-Context Pricing
Some models charge a higher rate for the whole request once your prompt passes a size limit, like a surcharge.
Definition
A pricing tier where a request whose input passes a token threshold is billed at higher rates for the whole request, not just the tokens past the threshold. In the site's catalog: Claude Haiku 5.5 above 100K input tokens (5x input and output); GPT-5.4, GPT-5.5, GPT-5.6, and GPT-6 models above 272K (2x input, 1.5x output); Gemini 3.1 Pro and Gemini 2.5 Pro above 200K (2x input, 1.5x output). Budget estimates for large contexts must account for the jump.
Example
On Claude Haiku 5.5, a 120K-token prompt crosses the 100K threshold, so input and output are both billed at 5x the base rate.