{T}
TokenMath.net
Original Research & IndicesIndustry Pricing Macroeconomics

AI Pricing Intelligence Reports

Data-backed research analyzing price-per-token macro trends, hardware amortization curves, prompt caching economics, and enterprise AI unit cost trajectories.

Featured Benchmark Report
September 15, 2026 · 9 min read

State of AI Token Pricing: September 2026 Benchmark Report

Comprehensive industry analysis across 30 frontier models: 44% annualized deflation in input token rates, widespread adoption of 90% prompt caching, and the emerging divide between reasoning deliberation and high-throughput utility inference.

Average input token price fell 44% year-over-year
Prompt caching read discounts averaged 82% across all 9 providers
Open-weights serverless inference reached an all-time low of $0.08/1M