Skip to content
AuraMatrix
← All posts

2026-11-16

LLM and GPU prices, week of November 9 to 15, 2026

No price moved on a vendor API or a cloud. 2 future changes were announced. The flagship token index closed at 100.0 (0.0%). 5 GPU models have an hourly reference price.

Model API price moves

No model changed price on a vendor’s own API or on a cloud this week.

Marketplaces showed 18 further listing changes. A marketplace’s price follows whichever provider it routes to and swings back and forth, so these are counted, not listed.

Announced, not yet in force

  • Gemini 3.8 Flash on Gemini Developer API: input $0.75 → $1.50, output $3.75 → $7.50, from 2027-01-01.
  • Gemini 4 Argon on Gemini Developer API: input $2.00 → $4.00, output $10.00 → $20.00, from 2026-12-01.

Models added

Claude Fable 5.2, Claude Mythos 5.2

Token price indices (trial)

Base date = 100. Official USD list prices, 3 parts input to 1 part output; a model joining or leaving never counts as a move.
TierEnd of weekChange in the weekModels
Flagship100.00.0%12
Mid100.00.0%4
Small100.00.0%7

GPU-hour reference prices (trial)

Median of each provider’s cheapest on-demand price, USD per GPU-hour. No change is given for a week in which the number of providers changed.
GPUEnd of weekChange in the weekLow to highProviders
NVIDIA B200$8.19—$5.98 – $16.117
NVIDIA H200$6.305—$3.59 – $10.607
NVIDIA H100$5.072—$1.99 – $10.988
NVIDIA A100 80GB$3.19—$1.19 – $5.0287
NVIDIA L40S$1.476—$0.79 – $2.254

Mainland China (CNY)

None of the recorded Alibaba Cloud pay-as-you-go GPU prices changed this week.

The week’s checks

10 daily checks ran this week, comparing published prices with their sources 1249 times; 1188 of them found the source still stating the recorded price. Day by day →

This report is computed entirely from the index’s records; no figure in it is typed in. List prices as published by each channel and provider, not invoices.

Free to quote and republish with the credit “Source: AuraMatrix (auramatrix.ai)”.