OpenRelay
NVIDIA and AMD GPUs by the hour, plus an OpenAI-compatible inference API for open-weight models
OpenRelay is the cheapest provider for MI300X and MI325X today, and under the market median on 11 of its 14 GPUs.
OpenRelay is a GPU cloud and inference API. Rent NVIDIA and AMD GPUs by the hour, from the RTX 3090 and RTX 4090 to the A100, H100 SXM, B200, B300 and AMD Instinct MI355X, as VMs with the GPU passed through or as bring-your-own-container pods. There is no commitment, quota request or reservation, and no egress fees; CPU, RAM and disk are included in the hourly rate. The same account calls open-weight LLMs (GPT-OSS, DeepSeek, GLM, Qwen, Gemma) through an OpenAI-compatible inference API billed per token. Based in San Francisco, backed by Y Combinator (S26).
At a glance · prices collected Nov 19, 2026
What's good about OpenRelay
- Support for multiple AI models
- Comprehensive documentation and API resources
- No commitment, quota request or reservation
- No egress fees; CPU, RAM and disk included in the hourly rate
OpenRelay GPU prices
14 GPU models across 28 configurations. Rank compares OpenRelay's cheapest on-demand listing for each GPU with every provider's cheapest.
| GPU | OpenRelay | -60%median+100% | Rank | vs. median | Report |
|---|---|---|---|---|---|
| H100 SXM 1× · community | $2.60/hr | #16 of 57 providers | -18% | ||
| RTX PRO 6000 1× · community | $1.55/hr | #10 of 41 providers | -22% | ||
| B200 1× · community | $6.30/hr | #14 of 36 providers | -17% | ||
| HGX B300 1× · community | $8.49/hr | #15 of 35 providers | -6% | ||
| RTX 5090 1× · community | $0.54/hr | #6 of 24 providers | -11% | ||
| RTX 4090 1× · community | $0.35/hr | #5 of 23 providers | -20% | ||
| L4 1× · community | $0.43/hr | #3 of 21 providers | -39% | ||
| H100 PCIe 1× · community | $2.20/hr | #3 of 15 providers | -20% | ||
| RTX 3090 1× · community | $0.18/hr | #7 of 14 providers | at median | ||
| RTX 4000 Ada 1× · community | $0.25/hr | #5 of 13 providers | -67% | ||
| MI300X 1× · community | $1.99/hr | Lowest of 11 providers | -33% | ||
| RTX 5000 Ada 1× · community | $0.50/hr | #3 of 5 providers | at median | ||
| MI355X 1× · community | $6.30/hr | #3 of 4 providers | +1% | ||
| MI325X 1× · community | $2.25/hr | Lowest of 2 providers | -26% |
All configurations
Prices last updated November 19, 2026
What a run costs on OpenRelay
24 hours at today's cheapest listing for each GPU, at the size it is listed.
1× H100 SXM for 24 hours
community at $2.60/GPU-hr
$62.40
$14.16 under the market median rate
1× RTX PRO 6000 for 24 hours
community at $1.55/GPU-hr
$37.20
$10.80 under the market median rate
1× B200 for 24 hours
community at $6.30/GPU-hr
$151.20
$31.08 under the market median rate
The market median is every provider's cheapest on-demand price for that GPU.
OpenRelay LLM API prices
Pay-per-token pricing, per 1M tokens.
Prices last updated November 19, 2026
| Model | Input/1M | Output/1M |
|---|---|---|
| $0.035 | $0.500 | |
| $0.045 | $0.450 | |
| $0.079 | $3.00 | |
| $0.240 | $8.40 |
OpenRelay pricing by model
Input, output, and batch rates, plus alternatives, for one model at a time.
Billing & terms
- On-demand GPUs
- Per GPU-hour, no commitment, quota request or reservation.
- Inference API
- Per million tokens, prepaid.
- Regions
- Locations are not published per listing. Ask sales@openrelay.inc for a specific region.
- Support
- Email support@openrelay.inc or sales@openrelay.inc. Documentation at docs.openrelay.inc.
Watch out for
- Data centre locations are not published
Features
- NVIDIA and AMD GPUs
- From the RTX 3090 and RTX 4090 to the A100, H100 SXM, B200, B300 and AMD Instinct MI355X.
- VMs or pods
- Rent a GPU as a VM with the GPU passed through, or run your own container as a pod.
- Per-token inference
- Open-weight LLMs (GPT-OSS, DeepSeek, GLM, Qwen, Gemma) billed per token, prepaid.
- API Integration
- Compatible with existing OpenAI SDKs and other client implementations.
Company
Published by OpenRelay to identify the company behind this listing.
- Legal entity
- OpenRelay Inc.
- Headquarters
- 🇺🇸 San Francisco, United States
- Contact
- sales@openrelay.inc
Getting started
Step 1
Create an Account
Sign up on the OpenRelay website to access services.
Step 2
Launch a GPU
Pick an NVIDIA or AMD GPU and start a VM, or run your container as a pod.
Step 3
Explore Inference APIs
Utilize various hosted models through the Inference API.
Step 4
Integrate API Key
Configure your application to authenticate with OpenRelay's API.
Step 5
Deploy Workloads
Use CLI tools or dashboard to manage deployments.
Where this page's facts come from
- GPU prices
- OpenRelay Public Pricing API (openrelay.inc) · last collected Nov 19, 2026
- LLM API prices
- openrelay.inc · last collected Nov 19, 2026
- Profile
- Reviewed Oct 6, 2026
- Method
- How we collect and check prices
Frequently Asked Questions
What GPU types does OpenRelay offer?
OpenRelay offers various GPU types including RTX 5000 Ada, RTX 5000 Ada, MI355X, MI355X, H100 SXM, H100 SXM, RTX 4000 Ada, RTX 4000 Ada, RTX 4090, RTX 4090, HGX B300, HGX B300, MI325X, MI325X, H100 PCIe, H100 PCIe, RTX 5090, RTX 5090, B200, B200, MI300X, MI300X, L4, L4, RTX PRO 6000, RTX PRO 6000, RTX 3090, RTX 3090. Check the pricing table above for current availability and pricing.
How do I get started with OpenRelay?
Create an Account, Launch a GPU, Explore Inference APIs, Integrate API Key, Deploy Workloads
What are OpenRelay's main advantages?
OpenRelay's main advantages include: Support for multiple AI models, Comprehensive documentation and API resources, No commitment, quota request or reservation, No egress fees; CPU, RAM and disk included in the hourly rate.
What are OpenRelay's limitations?
OpenRelay's main limitations include: Data centre locations are not published.
Compare Providers
Find the best prices for the same GPUs and models from other providers · All OpenRelay alternatives, ranked →
All OpenRelay Comparisons
The weekly price report
One email every Monday with the week's notable GPU and LLM API price moves, plus occasional product news. No spam.