Inference and training platform for open models: pay-per-token serverless APIs, GPU deployments, fine-tuning and reserved capacity.
Paid
hosted·ai
GPU-as-a-service platform that lets fleet owners, neoclouds and telcos resell GPUs as virtual GPU, bare metal and token services.
What is hosted·ai?
hosted·ai is a GPUaaS platform for companies that own or lease GPU capacity and want to sell it as AI cloud services. It bundles orchestration, multi-tenant isolation, scheduling and billing, so a provider can offer virtual GPU slices, dedicated cards, GPU plus virtual machines, bare metal and token-based inference endpoints from a single stack. Operators set overcommit ratios per GPU pool, meter customers by VRAM and TFLOPs consumed, and hand their own users a white-label portal and REST API. The software runs on Ubuntu with NVIDIA, AMD and other accelerators and connects to billing and customer systems such as WHMCS, HubSpot and Stripe. The vendor also runs GPU Mesh, a capacity network where owners can sell spare GPUs and other operators can source GPUs without buying hardware. Licensing is metered on VRAM: $0.0005 per GB-hour for VRAM managed and $0.0030 per GB-hour for VRAM consumed, with a $750 per month minimum, no up-front licence fee and no per-seat fees.
Use cases
- Resell GPU capacity as virtual GPU
- Sell token-based inference endpoints
- Launch a neocloud on owned GPUs
- Bill tenants by VRAM and TFLOPs
- Monetize idle GPU capacity
- Add GPUaaS to a telco or CSP cloud
Who it's for
Why use hosted·ai
- Sell several GPU products, from fractional virtual GPU to bare metal, from one platform instead of stitching tools together.
- Serve more customers per card through secure multi-tenancy and overcommit settings per pool.
- Usage-based licensing on VRAM with no per-seat fees, so cost follows the capacity you manage and sell.
- A white-label portal and REST API let you present the service under your own brand and plug it into existing billing.
hosted·ai features
- Multi-tenant GPU sharing with isolated tasks and memory isolation
- GPU pooling from low-cost shared to dedicated pools
- Automatic GPU scheduling to raise utilization
- Overcommit ratios set per GPU pool
- Billing by VRAM and TFLOPs consumed, instances, bare metal and services
- Unified management of KVM, Kubernetes and bare metal
- White-label self-service UI for admins and AI developers
- Full REST API for provisioning and usage data
- Role-based access control, audit logs and air-gap support
- GPU Mesh capacity network for selling or sourcing GPUs
- Packages that combine CPU, GPU, storage and network
hosted·ai integrations
How to get started with hosted·ai
- Install the platform on your GPU nodes, or connect it to your existing cloud platform.
- Create GPU pools, set overcommit ratios and define packages, regions and prices.
- Customers self-serve through your branded portal or the REST API and receive isolated GPU access.
- Usage is metered by VRAM, TFLOPs, tokens or instances and passed to your billing and customer systems.
hosted·ai limitations
- Built for providers that resell GPU capacity, not for end users who only need a GPU
- Runs on GPU hardware under Ubuntu
- Calculator figures are illustrative; a full quote comes from the vendor
- Licensing carries a minimum of $750 per month
hosted·ai pricing
hosted·ai has no free plan.
Prices can change and may vary by country or exclude tax. Check the official pricing page.
hosted·ai alternatives
The main hosted·ai alternatives listed on Debuly are CoreWeave, Lambda, Baseten, FlexAI, GPU.ai and MegaSpeed.Ai. The table compares pricing and platforms; order is A–Z.
| Tool | Pricing | Starting price | Platforms |
|---|---|---|---|
| hosted·ai | Paid | — | Web, API, Self-hosted |
| Baseten | Paid | — | Web, API, CLI, Self-hosted |
| CoreWeave | Paid | — | Web, API, Bare metal |
| FlexAI | Paid | $0.01, usage-based | API |
| GPU.ai | Paid | $0.24, usage-based | CLI, API |
| Lambda | Paid | $0.69, usage-based | Web, API, CLI |
| MegaSpeed.Ai | Paid | $0.24, usage-based | — |
Baseten
Inference platform for deploying open-source, custom and fine-tuned AI models, with per-token Model APIs and dedicated GPUs.
CoreWeave
AI cloud that rents NVIDIA GPU capacity by the hour, with on-demand, spot and reserved plans for model training and inference.
FlexAI
OpenAI-compatible inference API, dedicated GPU endpoints and fine-tuning for open models, billed per token or per GPU-hour.
GPU.ai
GPU cloud that routes each launch to the cheapest provider with stock, billed per second, for training and inference.
Lambda
GPU cloud for AI with on-demand NVIDIA instances and reserved 1-Click Clusters for training, fine-tuning and inference.
MegaSpeed.Ai
GPU and Kubernetes cloud for AI inference, rendering and pixel streaming, with usage-based pricing.
hosted·ai FAQ
How much does hosted·ai cost?
For providers selling on-premises or physical GPU, licensing is based on VRAM: $0.0005 per GB-hour for VRAM managed and $0.0030 per GB-hour for VRAM consumed, with a minimum of $750 per month. The vendor says there are no licensing fees to pay up front and no charges per seat, and a calculator on the pricing page gives illustrative estimates.
Does hosted·ai have a free plan?
No free plan is listed. The pricing page shows usage-based licensing with a $750 per month minimum and says subscribing to GPUs through GPU Mesh is free. Contact the vendor for a quote for your deployment.
Who is hosted·ai for?
It targets GPU fleet owners, neocloud operators, telcos, cloud service providers, energy providers and new GPU data center builders that want to sell GPU capacity rather than use it internally.
Does hosted·ai have an API or integrations?
Yes. It offers a full REST API and connects to billing and customer management systems including WHMCS, HubSpot and Stripe. A white-label UI is also included.
What hardware does hosted·ai run on?
The vendor says it runs on any GPU hardware under Ubuntu and works with NVIDIA, AMD and emerging accelerators.
Sources & verification
Facts on this page come from the official sources below, each with the date we read it. Written by Debuly; the vendor did not review this page.
| Source | Type | Accessed |
|---|---|---|
| hosted.ai/ | official | |
| hosted.ai/pricing | pricing | |
| docs.hosted.ai/user-panel/faqs/pricing-and-resource-policies-faqs | pricing | |
| hosted.ai/platform | official | |
| hosted.ai/ | official | |
| hosted.ai/glossary/vram-hour-billing | pricing |
Image credits
- Logo: from cdn.prod.website-files.com · fetched · Vendor trademark, used only to identify the product; removed on request
- Cover: none collected yet, so a generated tile is shown.
Spot an error? Report an issue with this listing. We also remove a logo or image on the owner's request.
More in Developer platforms 4
All developer platforms toolsFree browser calculators and an error debugger for PyTorch tensor shapes, plus a $29 one-time local toolkit called Tensor Preflight.
FreemiumSearch foundation model APIs: multilingual and multimodal embeddings, rerankers, document OCR and a Reader that turns URLs into Markdown.
FreemiumWebsite that compares LLM API token prices, context windows and benchmarks across 300+ models from many providers.
Free
