LLM Opex

Reports

LLM hosting prices: what changed in November 2026

Month in progress, updated 4 November 2026.

23 price cuts and 38 rises so far; the largest is 12% off DeepInfra nvidia/NVIDIA-Nemotron-3-Super-120B-A12B. Two models newly sold per token, six models withdrawn and 60 retirements this month.

23 price cuts · 38 price rises · 2 models newly sold per token · 6 withdrawn · 0 new GPU instances · 0 new regions

Biggest price moves

CloudWhatPriceChange
DeepInfranvidia/NVIDIA-Nemotron-3-Super-120B-A12B, priority, global$0.1275 → $0.1125 per 1M input tokens-11.8%
DeepInfranvidia/NVIDIA-Nemotron-3-Super-120B-A12B, flex, global$0.068 → $0.06 per 1M input tokens-11.8%
DeepInfranvidia/NVIDIA-Nemotron-3-Super-120B-A12B, batch, global$0.068 → $0.06 per 1M input tokens-11.8%
DeepInfranvidia/NVIDIA-Nemotron-3-Super-120B-A12B, standard, global$0.085 → $0.075 per 1M input tokens-11.8%
AzureStandard_NC4as_T4_v3, spot, westeurope$0.2925 → $0.3218 an hour+10%
AzureStandard_NC16as_T4_v3, spot, westeurope$0.6691 → $0.7359 an hour+10%
AzureStandard_NC8as_T4_v3, spot, westeurope$0.4179 → $0.4597 an hour+10%
AzureStandard_NC64as_T4_v3, spot, westeurope$2.42 → $2.66 an hour+10%
AzureStandard_NC4as_T4_v3, spot, eastus$0.1492 → $0.1641 an hour+10%
AzureStandard_NC64as_T4_v3, spot, eastus$1.23 → $1.36 an hour+10%

And 51 more.

Models newly sold per token

Withdrawn and retiring

The clouds' own announcements

The month before