Nvidia Plans Pricing
NVIDIA's developer API surface spans multiple commercial products with distinct pricing. build.nvidia.com offers free trial credits to call hosted NIM microservices for evaluation. NGC catalog downloads are free for development and prototyping. Production deployment of NIM/AI Enterprise requires a paid AI Enterprise license (per-GPU per-year) sold by NVIDIA sales or partners; cloud marketplace hourly rates are available on AWS, Azure, and GCP. CUDA, Omniverse, and DGX Cloud each have their own pricing tracks. Public list prices for AI Enterprise are not published.
Plans
Free trial credits to call hosted NIM microservices on build.nvidia.com for evaluation and prototyping. Account required.
- Free trial credits
- Hosted NIM microservices
- 100+ models (LLMs, retrieval, multimodal, healthcare)
Free downloads of NIM containers, pretrained models, and AI workflows from NGC for development and prototyping. Production use requires AI Enterprise license.
- NIM container downloads
- Pretrained models
- Workflow blueprints
- Development use only (production requires AI Enterprise)
90-day production trial license available upon request for AI Enterprise software stack including NIM, Triton, RAPIDS.
- 90-day production trial
- NIM microservices
- Triton Inference Server
- RAPIDS / cuOpt
Production AI Enterprise license, priced per GPU per year via NVIDIA sales or partners. Public list price not published; cloud marketplace hourly rates available on AWS / Azure / GCP.
- Production NIM microservices
- Enterprise support
- Per-GPU/year licensing
- Cloud marketplace hourly
- Triton Inference Server
- NVIDIA Developer API
NVIDIA's hosted GPU cloud service. Pricing per-instance-hour or per-month commit, billed directly or via cloud marketplaces.
- Multi-cloud GPU access
- Pre-configured AI training stack
- Per-instance-hour or commit pricing