Deploy GPU compute and model endpoints
Call a model. Deploy a model. Rent a GPU.
OneInfer Cloud gives you three ways to run a workload from one account: call catalog models through the shared Model APIs, deploy a public, private or fine-tuned model as a model deployment, or rent a GPU instance by the hour for training, development or custom serving.
Route between endpoints with a router, and keep model files and datasets in regional storage. API keys, credits and usage stay in one Console.
Where each workload runs
- Prototype a chat featureShared, usage-based access to catalog models.
- Serve a fine-tuned modelYour model on GPU workers you configure.
- Train or run custom codeA GPU machine by the hour, with SSH access.
Where each workload runs. Prototype a chat feature: Model API (shared). Shared, usage-based access to catalog models. Serve a fine-tuned model: Dedicated endpoint. Your model on GPU workers you configure. Train or run custom code: GPU instance. A GPU machine by the hour, with SSH access.
What you can deploy with OneInfer Cloud
On-demand GPU instances
Compare available GPU types, providers, regions, VRAM, and hourly prices. Select the capacity that fits training, development, or custom inference workloads.
Dedicated model endpoints
Deploy a supported public, private, or fine-tuned model with selected compute, region, worker limits, request concurrency, and scaling configuration.
Intelligent API endpoints
Attach dedicated or serverless endpoints behind one API endpoint and configure routing for the models and infrastructure used by your application.
Cloud deployment capabilities available in OneInfer
GPU marketplace
Search and filter available GPU capacity by GPU model, provider, region, memory, availability, and price before creating an instance.
Configurable endpoints
Choose a model template, GPU, provider, region, minimum and maximum workers, scaling mode, and concurrency for each deployment.
Persistent storage
Create and manage regional storage volumes for model files, datasets, checkpoints, and workloads that need data beyond an instance lifecycle.
Instance lifecycle controls
Create, inspect, start, stop, and delete supported GPU instances from the console, with instance state and connection details kept in one place.
OneInfer API access
Manage API keys in the same account and direct requests to the model or endpoint selected for your application.
Usage and credit visibility
Keep deployment access, account credits, API usage, and infrastructure controls together instead of maintaining separate provider accounts and billing views.