| <html lang="en"> | |
| <head> | |
| <meta charset="utf-8"> | |
| <meta name="viewport" content="width=device-width, initial-scale=1"> | |
| <meta name="description" content="NVIDIA Dynamo open-source distributed inference framework."> | |
| <title>NVIDIA Dynamo</title> | |
| <link rel="stylesheet" href="style.css"> | |
| </head> | |
| <body> | |
| <main> | |
| <h1>NVIDIA Dynamo</h1> | |
| <p><strong>Open-source, low-latency inference for generative AI.</strong></p> | |
| <p>NVIDIA Dynamo is a modular inference framework for serving generative AI models in distributed environments. It scales workloads across GPU fleets with intelligent resource scheduling, request routing, optimized memory management, and data transfer.</p> | |
| <img src="https://huggingface.co/spaces/ai-dynamo/README/resolve/main/nvidia-dynamo-inference-architecture.png" alt="NVIDIA Dynamo architecture and components" style="display: block; width: 100%; height: auto; margin: 24px 0;"> | |
| <p><a href="https://github.com/ai-dynamo/dynamo" target="_blank" rel="noopener noreferrer">GitHub repository</a> · <a href="https://docs.nvidia.com/dynamo/latest/" target="_blank" rel="noopener noreferrer">Documentation</a></p> | |
| <hr> | |
| <h2>Built for production inference</h2> | |
| <ul> | |
| <li><strong>Distributed serving:</strong> Deploy and scale inference across GPU fleets.</li> | |
| <li><strong>Disaggregated inference:</strong> Independently optimize prefill and decode workloads.</li> | |
| <li><strong>Open ecosystem:</strong> Works with SGLang, TensorRT-LLM, and vLLM.</li> | |
| </ul> | |
| <p><strong>Datasets and evaluation:</strong> This organization publishes Dynamo-related datasets and reproducible evaluation fixtures.</p> | |
| <p>More source code, examples, and project updates: <a href="https://github.com/ai-dynamo" target="_blank" rel="noopener noreferrer">github.com/ai-dynamo</a></p> | |
| </main> | |
| </body> | |
| </html> | |