GPU workload page

H100 NVL batch inference price by provider

Batch inference can use cheaper and less stable pricing modes when queues are flexible and retries are acceptable. For H100 NVL, the current tracked provider floor is $2.27/hr on Vast.ai.

94GB HBM3 batch inference 24GB+ floor Updated Jun 21, 2026
Cheapest provider
Vast.ai
$2.27/hr
Monthly estimate
$1,656/mo
730 hours at current median price
Provider coverage
3
6 current rows updated Jun 21, 2026
Market gap
+$2.14/hr
Floor is GCP L40

Current H100 NVL rows for batch inference

H100 NVL clears the 24GB+ filter for batch inference. The broader batch inference floor is $0.13/hr with L40 on GCP, so the cheapest H100 NVL row is $2.14/hr above that floor.

Updated Jun 21, 2026
Provider Type Median hourly Monthly Workload page Offers Updated
Vast.ai on-demand $2.27/hr $1,656/mo Vast.ai batch inference 9 Jun 21, 2026
RunPod spot $2.59/hr $1,891/mo RunPod batch inference 1 Jun 21, 2026
RunPod community $2.59/hr $1,891/mo RunPod batch inference 1 Jun 21, 2026
RunPod on-demand $3.19/hr $2,329/mo RunPod batch inference 1 Jun 21, 2026
Azure spot $6.28/hr $4,586/mo Azure batch inference 6 Jun 21, 2026
Azure on-demand $6.98/hr $5,095/mo Azure batch inference 6 Jun 21, 2026

H100 NVL for batch inference: current provider pricing

This page starts with the GPU instead of the provider, then filters the live market to rows that match batch inference. It is useful when you have already chosen H100 NVL and need the cheapest provider path for the workload.

Offline inference queues, embedding jobs, catch-up workloads, and cost-sensitive internal processing. The table keeps provider/GPU and provider/workload links close together so you can keep narrowing the shortlist without going back to the homepage.

Cheapest provider right now

Cheapest H100 NVL provider for batch inference

The cheapest tracked H100 NVL row for batch inference is $2.27/hr on Vast.ai with on-demand pricing. The broader batch inference floor is $0.13/hr with L40 on GCP, so the cheapest H100 NVL row is $2.14/hr above that floor.

Methodology and freshness

How this GPU workload page is computed

We filter the compare payload to H100 NVL, require 24GB+ workload fit, and sort current spot, community, and on-demand provider rows by median hourly price.

H100 NVL batch inference pricing FAQ

Is H100 NVL good for batch inference?

H100 NVL clears the 24GB+ filter for batch inference. The cheapest tracked H100 NVL row for batch inference is $2.27/hr on Vast.ai with on-demand pricing.

What is the cheapest H100 NVL provider for batch inference?

The cheapest tracked H100 NVL row for batch inference is $2.27/hr on Vast.ai with on-demand pricing.

How does H100 NVL compare with the broader batch inference market?

The broader batch inference floor is $0.13/hr with L40 on GCP, so the cheapest H100 NVL row is $2.14/hr above that floor.

How fresh is this H100 NVL batch inference page?

The page is recalculated from the latest stored spot, community, and on-demand rows. The freshest qualifying row visible here is from Jun 21, 2026.