Serve Qwen3.8-2.4T-A95B, a 2.4T-Parameter Model, with Configurable Reasoning on NVIDIA GB300 NVL72
Source
developer.nvidia.com
Author
Michelle Horton
Date
Context
Alibaba's Qwen3.8-2.4T-A95B distributes its capacity across many specialized model components, activating 95 billion of its 2.4 trillion parameters for each text unit it processes. NVIDIA's deployment article describes serving it on a GB300 NVL72 system containing 72 GPUs.
Developers can also select low, high or extra-high reasoning effort for each request, choosing how much computation to devote to an answer. These controls offer flexibility within a large deployment: activating only part of the model does not make the full model a single-GPU workload.
Terms in this piece · Glossary
open weights — A model whose trained parameters are published for anyone to download and run — unlike API-only models you can access but never possess.
token — The chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.
