Voltage Park · Rate Limits

Voltage Park Rate Limits

The Voltage Park On-Demand API does not publish explicit per-second or per-minute request rate limits in its public documentation. Practical limits are governed by GPU inventory and account quotas: on-demand capacity ranges from 1 up to ~1016 GPUs, bare-metal rentals are provisioned in multiples of eight, and reserved capacity spans 32 to 8000+ GPUs. API tokens are bearer-scoped per organization and expire after one year. Specific HTTP request throttling values are not reconciled in this artifact.

Voltage Park Rate Limits is the machine-readable rate-limit profile for Voltage Park on the APIs.io network, conforming to the API Commons Rate Limits specification.

It captures 5 rate-limit definitions, measuring requests, gpus, and days.

The profile also includes 2 backoff/retry policies defined and response codes documented for throttled.

Tagged areas include GPU, Cloud, AI Infrastructure, H100, and H200.

5 Limits Throttle: 429
GPUCloudAI InfrastructureH100H200Bare MetalRate LimitingQuotasThrottling

Limits

API Request Rate token
requests
not publicly documented
No explicit RPM/RPS published; bearer token scoped to one organization.
On-Demand GPU Capacity account
gpus
up to ~1016 per on-demand cluster
Deployments are bounded by available inventory at the chosen location.
Bare-Metal Node Granularity account
gpus
multiples of 8
gpu_count on bare-metal provisioning must be a positive multiple of eight.
Reserved Capacity account
gpus
32 to 8000+
Reserved clusters require 6+ month commitments and are sized by contract.
API Token Lifetime token
days
365
Bearer API tokens are valid for one year before expiring.

Policies

Capacity-Bounded Provisioning
Requests to deploy fail when GPU inventory is unavailable at the selected location or host node.
Backoff Strategy
Clients should implement exponential backoff with jitter and honor Retry-After on 429 responses.

Sources