NVIDIA announced Dynamo, a disaggregated AI workload orchestration software, and open sourced it.

True if, by 2025-09-29, NVIDIA had publicly announced software called Dynamo, described as an inference/AI workload orchestration framework built around disaggregated serving (e.g. separating prefill and decode across GPUs), and had released its source code under an open-source licence in a public repository. Mostly true if both happened but a minor detail is off (e.g. the announcement described it mainly as an 'inference framework' rather than 'orchestration', or only most components were open-sourced at release). Mixed if Dynamo was announced but only partly open-sourced (e.g. key components kept proprietary) as of that date. Mostly false if Dynamo exists but was not open-sourced by 2025-09-29, or is not about disaggregated workloads. False if NVIDIA announced no such product, or records show Dynamo is closed-source.

Set before the research began.

1. E1 NVIDIA Dynamo Open-Source Library Accelerates and Scales AI Reasoning Models NVIDIA Newsroom · 18 March 2025 Own record

   “NVIDIA Dynamo is fully open source and supports PyTorch, SGLang, NVIDIA TensorRT™-LLM and vLLM to allow enterprises, startups and researchers to develop and optimize ways to serve AI models across disaggregated inference.”

2. E2 NVIDIA Dynamo Open-Source Library Accelerates and Scales AI Reasoning Models NVIDIA Newsroom · 18 March 2025 Own record

   “It orchestrates and accelerates inference communication across thousands of GPUs, and uses disaggregated serving to separate the processing and generation phases of large language models (LLMs) on different GPUs.”

3. E3 GitHub - ai-dynamo/dynamo: A Datacenter Scale Distributed Inference Serving Framework GitHub (repository under the ai-dynamo organisation) Own record

   “Dynamo is the orchestration layer above inference engines — it doesn't replace SGLang, TensorRT-LLM, or vLLM, it turns them into a coordinated multi-node inference system.”

4. E4 China's Ant Group raises red flags on US tech and open-source AI practices TechRadar Pro · 16 September 2025

   “For example, Nvidia’s Dynamo (open-sourced in March 2024) is marketed as the operating system of AI, however it’s been specifically optimized for Nvidia GPUs.”

NVIDIA's dated official announcement of 2025-03-18 introduced Dynamo as "fully open source" software for serving AI models across disaggregated inference (E1). The same announcement says Dynamo orchestrates inference across GPUs and uses disaggregated serving to run the processing and generation phases on different GPUs (E2). A public GitHub repository describes Dynamo as an orchestration layer for multi-node inference (E3), and an independent TechRadar report from 2025-09-16 says Dynamo had been open-sourced (E4). TechRadar gives the year as 2024 rather than 2025, and the fetched text does not show the licence type, though NVIDIA's announcement describes Dynamo as 'fully open source' [E1] and the code is in a public repository [E3].