Welcome to InstantGPU
Welcome to the InstantGPU Documentation.
InstantGPU provides flexible GPU infrastructure for developers, AI engineers, startups, and teams building modern AI applications. Whether you're experimenting with a new model, running GPU-intensive workloads, deploying AI inferenceInference is running live trained machine learning models to generate outputs, predictions, or tokens with ultra-low latency., or building an application for production, InstantGPU gives you the compute infrastructure needed to get started quickly.
With InstantGPU, you can access powerful GPU resources without having to build and maintain complex GPU infrastructure from scratch. Choose the infrastructure model that fits your workload, configure your environment, deploy your application, and start running your AI workloads.
Our documentation is designed to help you understand the InstantGPU platform and get the most out of your GPU resources. From your first deployment to production integrations, you'll find practical information, configuration instructions, API documentation, and troubleshooting resources throughout this documentation.
What Can You Build with InstantGPU? #
InstantGPU can power a wide range of AI and GPU-intensive applications:
AI Inference
Run trained AI and machine learning models on high-performance GPUs with dedicated vLLM and TensorRT engines.
LLM Applications
Build chatbots, AI assistants, text-generation applications, and other LLM-powered products with zero setup.
Image Generation
Run image generation, editing, upscaling, and computer-vision workloads using Stable Diffusion and Flux pipelines.
Video Generation
Power AI video generation, frame interpolation, and other GPU-intensive video workloads at scale.
Audio & Speech
Run speech recognition, transcription, text-to-speech, and audio-processing workloads via production endpoints.
Embeddings & RAG
Build semantic search, document retrieval, knowledge-base, and RAG applications with sub-50ms vectorization.
Custom AI Workloads
Deploy your own custom models, frameworks, Docker containers, dependencies, and complex GPU applications.
Choose the Right Way to Run Your Workload #
Different AI applications require different infrastructure setups. InstantGPU provides multiple ways to run your workloads depending on your requirements for control, scalability, and usage:
Use dedicated GPU infrastructure when you need full control over your environment, persistent workloads, or a custom VM.
Use serverless infrastructure when you need workloads to auto-scale with demand, scale to zero, and charge only per second.
Use ready-to-integrate AI APIs when you want to call frontier models directly without managing the underlying weights or servers.
Explore InstantGPU Documentation #
Use the documentation navigation to learn about the platform, deploy workloads, integrate APIs, and solve common issues:
Getting Started
Learn the basics of InstantGPU and launch your first GPU workload in 5 steps.
GPU Cloud
Learn how to deploy, SSH into, and manage dedicated GPU infrastructure.
Serverless GPU
Learn how to run scalable AI workloads without managing traditional servers.
AI Endpoints
Connect your applications with production-ready AI APIs for vision, speech & LLMs.
API Reference
Find technical details and SDKs for integrating InstantGPU APIs into your code.
GPU Templates
Explore preconfigured environments for PyTorch, vLLM, ComfyUI, and training stacks.
Guides
Follow practical tutorials for deploying and running common production AI workloads.
Troubleshooting
Find instant solutions to common GPU, deployment, API, and connection issues.
Start Building with InstantGPU #
New to InstantGPU?
Start with the Getting Started Guide to understand the platform and launch your first workload.
Already familiar with GPU infrastructure?
Use the documentation navigation on the left or search above to jump directly to the section you need.