Build and run AI models with fast deployment
SynapsAI Cloud is a managed GPU platform for deploying Hugging Face models with ultra-low load times, predictable billing, and OpenAI-compatible APIs — without operating Kubernetes, VMs, or inference servers yourself. We are in public beta and actively improving the platform. Share feedback via feedback and feature requests.Quickstart
Deploy a model and run your first API call in minutes.
API reference
Authentication, endpoints, and OpenAI compatibility.
Why SynapsAI Cloud?
Sub-second model loading
Prepped artifacts and local NVMe storage deliver load times measured in seconds, not minutes.
Secure multi-tenant infrastructure
Models run on shared GPU infrastructure with sandboxing, filesystem isolation, and reserved compute resources.
Transparent pricing
Memory-based compute pricing, usage dashboards, and cost controls built for teams.
How it works
1
Connect your model
Point SynapsAI at almost any Hugging Face repository. If the model is not on the Hub, upload an artifact and select it when you deploy.
2
Set scale
Configure minimum and maximum instances, autoscaling thresholds, and worker timeout.
3
Call the API
Use the Python SDK, cURL, or any OpenAI-compatible library against your private endpoint.
Documentation
Core concepts
Deployment types, pricing, credits, and model lifecycle.
Deploy a model
Requirements, precision, quantization, and scaling.
Inference quickstart
SDK setup, authentication, and your first request.
Migrate from OpenAI
Switch existing OpenAI integrations with minimal changes.
Knowledge bases
Index documents and search them through the vector store API.
Agents
Run a saved agent with tools, MCP, and knowledge bases.
Examples
Code samples for every supported pipeline task.
Supported tasks
Pipeline names, endpoints, and streaming support.
Manage models
Configuration, analytics, logs, and lifecycle states.
Integrations
LangChain, LlamaIndex, FAISS, Gradio, and FastAPI.
Need help? Check Troubleshooting, the status page, or contact support.