We are at re:Invent 2026Book my merch

Vast.ai

Every Vast.ai GPU-hour, accounted for.

Cloud Intelligence™ imports your Vast.ai charges every six hours: rented GPU instances and serverless endpoints, by GPU model, component, instance and location. Right next to the rest of your AI and cloud spend.

Read-only API key · Refreshed every 6 hours · Last 4 days reloaded

Vast.ai · October
$62,480
GPU time84%Storage11%Bandwidth5%
GPU-hours
31,200
Instances
148
Serverless share
32%
By GPU model and component
H100 SXM GPU time$22,140
A100 80GB GPU time$13,480
RTX 4090 GPU time$9,860
Storage instance disk$6,870
L40S GPU time$5,920

Trusted by teams training, fine-tuning, and serving AI models in production

Akamai
Blumira
Chess.com
Current
OneFootball
Promptly
Recurly
Lusha
Mux

what we import

Vast.ai data, in the dimensions you already use.

Every charge maps to standard Cloud Analytics dimensions, including workload and region, plus Vast.ai labels for charge type, component, instance, serverless endpoint, GPU model, location and template.

Cloud AnalyticsVast.aiExample
ProviderVast.aiVast.ai
Billing accountThe account name you enteredacme-ml
WorkloadThe serverless endpoint, or the rented instancellm-serve-prod
ServiceServerless Inference, or GPU InstanceGPU Instance
SKUThe billed component, with the GPU model on GPU timeGPU time · H100 SXM
RegionCountry of the host machineUnited States
Plus Vast.ai labels
Charge typeComponentInstanceServerless endpointGPU modelLocationTemplate
Refreshed every 6 hours · last 4 days reloaded each time

what you can do with it

From the Vast.ai bill to a decision

Cost by workload

Slice by instance, serverless endpoint, GPU model, component or template.

Serverless vs rented

See what share of spend runs on serverless endpoints.

GPU-hours and cost

Rented instances and serverless endpoints, GPU-hours and cost side by side.

GenAI Intelligence

Vast.ai on the GenAI dashboards, next to your other AI providers.

Budgets and alerts

Guardrails on GPU spend, with alerts to your team.

Allocations

Put Vast.ai spend on teams and products, by instance or endpoint.

read-only access

Connect Vast.ai in three steps

1

Create an API key.

In the Vast.ai console, with read access to billing and instances.

2

Connect in the catalog.

Enter an account name and the key, then test the connection.

3

Costs refresh every 6 hours.

We reload the last four days each time, by UTC day.

No write permissions needed. Connect each Vast.ai account separately.

Start free

Frequently asked
questions

How do I connect Vast.ai to Cloud Intelligence™?

Create an API key in the Vast.ai console with read access to billing and instances. Then open Data ingestion and integrations > Integrations, find Vast.ai, enter an account name and the key, and test the connection before saving. You need the Billing Profiles Admin permission, and each Vast.ai account is connected separately.

How fresh is Vast.ai data?

We read your Vast.ai charges every six hours and reload the last four days each time. Vast.ai bills against the UTC day, so the current day's cost rises until that day closes and can still change shortly after midnight UTC.

Which dimensions and labels are imported?

Charges map to Provider, Billing account, Workload, Service, SKU and Region, with each instance's GPU time, storage, download and upload bandwidth as separate daily lines. Labels include Charge type, Component, Instance, Serverless endpoint, GPU model, Location and the instance template name.

Are Vast.ai credit top-ups included?

No. A credit top-up is a prepaid balance rather than consumption, so counting it as cost would double-count the spend it later pays for.

Why does a cost line have no GPU model or location?

GPU model, host machine and location come from Vast.ai's list of your current instances, which leaves out destroyed instances. An instance created and destroyed within a few hours is still billed in full, but may be reported without a GPU model or location.

How does Vast.ai show up next to my other AI providers?

All Vast.ai spend is reported as GenAI spend and appears on the GenAI dashboards, and the Charge type label separates serverless endpoints from rented instances. Vast.ai's billing data has no model names or token counts, so its spend is not broken down by model or by input and output tokens.

Connect Vast.ai. Know what every GPU-hour costs.

Free to start. Read-only. Refreshed every 6 hours.