Runpod Serverless Flex pricing: why VRAM alone is the wrong comparison
Runpod's Flex tiers range from $0.58/hour 16GB classes to high-end B300 capacity. The useful decision is model fit, hardware class, and measured cost per job—not VRAM alone.
← Back to GPU Cloud
✦ NEW DEAL
Together AI · GPU Cloud · GPU Instance
Together GPU Clusters — B200 preemptible
Together AI GPU Clusters list preemptible NVIDIA HGX B200 capacity at $4.09 per GPU-hour versus $8.19 per GPU-hour on-demand. Preemptible nodes are available on Kubernetes clusters in all regions, can be reclaimed with up to a five-minute drain window, and are billed sub-hourly.
DEAL SCORE87Current-offer attractiveness
CURRENT PRICE$4.09 / hrNative listed value
Pricing
Pricing details
Secondary price terms only; primary pricing stays in the deal summary.
BILLINGOther
Fit & constraints
Fit & constraints
Use-case fit and limitations from the existing resource record.
AVAILABILITY
Available
CONTRACT
Preemptible; sub-hourly metering every 1–2 minutes; up to 5-minute drain window on reclamation.
Resource facts
Key specs
Non-price resource facts only; pricing, history, and evidence stay in their own sections.
GPU MODELNVIDIA HGX B200
GPU COUNT1
VRAM180 GB
LOCATIONAll Together GPU Cluster regions
Trust layer
Verified evidence
Decision visuals remain traceable to the existing source and verification record.
Independent ranking · native prices · evidence-linked · commercial relationships isolated from scoring and publication.
WATCH