Falling token prices obscure a surge in tracked inference demand, which hit 206.5T tokens/week as of 2026-11-10 after rising +33% over 30 days and +778% since 2026-05-29, lifting the demand index to 878.1. The market is expanding, not imploding: open-source models handled 75.5% of volume at a demand-weighted effective price of $1.07 per 1M tokens, while OpenAI took 8%, Claude 7%, and Google 3%, proving cheap distribution fuels booming consumption.
Word count:
Falling(1) token(2) prices(3) obscure(4) a(5) surge(6) in(7) tracked(8) inference(9) demand,(10) which(11) hit(12) 206.5T(13) tokens/week(14) as(15) of(16) 2026-11-10(17) after(18) rising(19) +33%(20) over(21) 30(22) days(23) and(24) +778%(25) since(26) 2026-05-29,(27) lifting(28) the(29) demand(30) index(31) to(32) 878.1.(33) The(34) market(35) is(36) expanding,(37) not(38) imploding:(39) open-source(40) models(41) handled(42)… It is the output-side companion to the Compute Tightness Index (the GPU input).
884.9Demand index (100 = 2026-05-31)
208.1TTokens / week
▲ +29%Demand, 30 days
75.7%Open-source share
$1.07Effective $/1M
$221.7MImplied spend/week*
As of 2026-11-11, tracked AI inference runs at 208.1T tokens/week — up +29% in 30 days and +785% since 2026-05-06. Open-source carries 75.7% of those tokens, holding the demand-weighted price near $1.07/1M — a fraction of frontier list prices even as frontier models themselves got pricier.
Open vs closed-model pricing
Blended price per million tokens. The closed series demand-weights each named provider group's median price by its trailing-seven-day token volume.
$0.45Open-weight models · $/1M tokens · 75.7% of tracked demand
$3.82Closed models · $/1M tokens · 17.7% of tracked demand
Open-weight $/1MClosed-model $/1M
Open-weight = the Open-source provider group. Closed = Claude, OpenAI, Google, and xAI. The unclassified “Other” bucket is excluded from the closed-model price.
Built from OpenRouter's model rankings + pricing, grouped into Claude / OpenAI / Google / xAI / Open-source / Other. Token counts are OpenRouter's trailing-7-day figures (the rankings "week" view) — i.e. each point is tokens processed over the prior week, not a single day. Per group: blended median $/1M (75/25 input/output) and weekly total_tokens. Demand index = latest weekly total ÷ the first week's total × 100. Effective $/1M = demand-weighted blended price (as-of prices carried forward across snapshots where only one side updated). Series since 2026-05-31; it grows as new snapshots land. Full history as CSV. Coverage is OpenRouter traffic — a large but partial slice of the market.
Pair it with the Compute Tightness Index (GPU input) for both sides of the AI-compute economy. Query the raw data from an agent via the Bargo MCP endpoint (get_inference_economics).