Breaking

Silicon Data LLM Token Index Falls to 97 Cents, Below Half Its May Peak

September 1, 2026 at 17:47 EDT

Foundation Models
Sales & Business

Silicon Data’s LLM Token Expenditure Index closed at 97 cents per million tokens on Monday, its first sub-dollar print since the December launch and less than half the $2.07 high recorded on May 28. The broad-market series, published on Bloomberg as SDLLMTK, also dropped 8.6% over the prior seven days, extending a months-long slide in the usage-weighted cost of large language model inference.

September 1, 2026 · Silicon Data

LLM token costs drop below $1

The LLM Token Expenditure Index (SDLLMTK) closed at 97 cents per million tokens on Monday — the first sub-dollar print since the December launch, and less than half the $2.07 high on May 28.

Peak vs Monday close (USD per million tokens)

$2.07

May 28 peak

$0.97

Monday close

Column height is proportional to the index level. Monday’s print is less than half the May high.

-8.6%

7-day change

<$1

First sub-dollar close

80%

GPT-5.6 Luna July cut

Buyers

Cheaper tokens and routing toward lower-cost models, including open-weight hosts, reduce the effective cost of inference as usage still expands.

Labs and tasks

Token deflation can squeeze revenue against fixed compute spend. More tokens per task can still raise the cost of completed work even as the unit price falls.

What can move the index

Pricing (provider rate-card cuts) → Mix (traffic shifting to cheaper models) → Composition (models entering or leaving the usage-weighted basket). The reading is a market baseline, not any customer’s invoice.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.