CarbonWatch

Generative search engines

A question, an answer written from pages fetched and cited. The model reads far more than it writes, but only the answer crosses the network towards you.

Billed on volume received, at 2.5 Wh per thousand tokens. Fetching the pages happens at the provider's end and appears nowhere in this estimate.

ServiceInference regionIntensity500-token answer
PerplexityUnited States380 g/kWh0.48 g · 500 tokens
You.comUnited States380 g/kWh0.48 g · 500 tokens
PhindUnited States380 g/kWh0.48 g · 500 tokens