OpenRouter: GPT-5.6 Promo Sparks 13.8x Token Surge

OpenRouter cut prices on GPT-5.6 Terra and Luna models and recorded a 13.8x jump in token consumption. The move demonstrates Jevons paradox in action: lower inference costs shift workflows from manual to agent-driven, expanding total compute demand rather than shrinking it. Support teams now run extra validation passes, founders scan every customer call, and developers request multiple iterations per task. This pattern directly counters claims that cheaper models will slow AI data center buildouts, as more tasks become economically viable for offload to AI.