From the archive
NVIDIA makes NIM inference containers available
Packaged model-serving containers offer a deployment route across workstations, data centers, and cloud infrastructure.
NVIDIA announced availability of NIM inference microservices, which package AI models as optimized containers. The offering gives applications a deployment route across supported workstations, data centers, and cloud infrastructure.
A packaged serving environment
NIM combines the model with the software used to serve it on NVIDIA hardware. That reduces the amount of separate setup involved in bringing together model weights, runtime components, and an application endpoint.
The announcement includes language, image, speech, and other model services, with more than 40 NVIDIA and community models available to try through its catalog.
Development and production access
The online microservices were available for experiments without charge. Production deployment was offered through NVIDIA AI Enterprise. Free access for Developer Program members to download and test NIM on their own infrastructure was announced for the following month.
The NVIDIA profile covers its model-serving software, enterprise licensing, and hardware-related deployment options.