AI companies are shifting competition from model capability alone toward price-performance, with falling costs driving sharply higher usage. OpenAI says its GPT-6 Sol and GPT-6 Luna releases will reduce prices for top business customers by 50% from previous iterations, while Anthropic says Opus 5.5 costs about 40% less to run than Opus 5 while retaining top intelligence. Elon Musk's xAI released Grok 4.7 around a price-performance pitch, although competing releases eroded much of its advantage within 24 hours. The trend poses margin and revenue-growth challenges for OpenAI and Anthropic, whose valuations are measured in multiple trillions of dollars, but it could expand the overall AI market by making usage more affordable. Citadel Securities said falling per-token costs are increasing usage and overall AI spending, while Morgan Stanley views competition from Chinese providers as another positive signal for demand for computing capacity. The dynamic reflects Jevons Paradox, the economic theory that lower resource costs can increase consumption. DeepSeek's V4.1 Flash, an open-weight model (whose parameters are available for users to run or adapt), rose to number one on OpenRouter's leaderboard after usage jumped 172% this week. The broader competition is pushing leading AI labs toward cheaper models, making earlier concerns about AI price shock appear less significant.