specification: API Commons Plans
specificationVersion: '0.1'
schema: https://raw.githubusercontent.com/api-evangelist/interface-research/main/schema/api-commons.yml#/$defs/Plans
provider: Nvidia
providerId: nvidia
created: '2026-05-04'
# Provenance stamped 2026-08-11: this artifact was written by the API Evangelist
# bulk sweep dated 2026-05-04, not harvested from the provider. See roadmap#35.
method: generated
modified: '2026-05-05'
reconciled: false
tags:
  - GPU
  - AI
  - Machine Learning
  - Computing
  - Graphics
description: 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.
notes: NVIDIA does not publish list pricing for AI Enterprise per-GPU/year on its public website. Free-tier
  credit values for build.nvidia.com fluctuate. Update once NVIDIA publishes specific list rates.
sources:
  - https://www.nvidia.com/en-us/data-center/products/ai-enterprise/
  - https://build.nvidia.com/
  - https://docs.api.nvidia.com/
  - https://catalog.ngc.nvidia.com/
plans:
  - id: nvidia-build-trial
    name: build.nvidia.com Free Credits
    type: freemium
    description: Free trial credits to call hosted NIM microservices on build.nvidia.com for evaluation
      and prototyping. Account required.
    entries:
      - label: Trial credits
        name: build_credits
        type: quota
        metric: credit
        limit: see build.nvidia.com (rotating free-credit allocation)
        timeFrame: usage
        geo: global
        domains:
          - https://integrate.api.nvidia.com
          - https://api.nvcf.nvidia.com
        unit: 1
        price: '0.00'
        userMultiplied: false
    elements:
      - name: Free trial credits
      - name: Hosted NIM microservices
      - name: 100+ models (LLMs, retrieval, multimodal, healthcare)
  - id: nvidia-ngc-catalog
    name: NGC Catalog (Development)
    type: freemium
    description: Free downloads of NIM containers, pretrained models, and AI workflows from NGC for development
      and prototyping. Production use requires AI Enterprise license.
    entries:
      - label: NGC catalog downloads
        name: ngc_downloads
        type: metered
        metric: download
        limit: -1
        timeFrame: usage
        geo: global
        unit: 1
        price: '0.00'
        userMultiplied: false
    elements:
      - name: NIM container downloads
      - name: Pretrained models
      - name: Workflow blueprints
      - name: Development use only (production requires AI Enterprise)
  - id: nvidia-ai-enterprise-trial
    name: NVIDIA AI Enterprise — 90-Day Trial
    type: freemium
    description: 90-day production trial license available upon request for AI Enterprise software stack
      including NIM, Triton, RAPIDS.
    entries:
      - label: 90-day trial
        name: aie_trial
        type: subscription
        metric: contract
        limit: 1
        timeFrame: usage
        geo: global
        unit: 1
        price: '0.00'
        userMultiplied: false
    elements:
      - name: 90-day production trial
      - name: NIM microservices
      - name: Triton Inference Server
      - name: RAPIDS / cuOpt
  - id: nvidia-ai-enterprise
    name: NVIDIA AI Enterprise
    type: enterprise
    description: 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.
    entries:
      - label: Per-GPU subscription
        name: aie_gpu_year
        type: subscription
        metric: gpu
        limit: -1
        timeFrame: year
        geo: global
        unit: 1
        price: contact sales
        userMultiplied: false
      - label: Cloud marketplace hourly
        name: aie_hourly_cloud
        type: metered
        metric: gpu-hour
        limit: -1
        timeFrame: usage
        geo: global
        unit: 1
        price: see AWS/Azure/GCP marketplace
        userMultiplied: false
    elements:
      - name: Production NIM microservices
      - name: Enterprise support
      - name: Per-GPU/year licensing
      - name: Cloud marketplace hourly
      - name: Triton Inference Server
      - name: NVIDIA Developer API
  - id: nvidia-dgx-cloud
    name: DGX Cloud / DGX Cloud Lepton
    type: usage-based
    description: NVIDIA's hosted GPU cloud service. Pricing per-instance-hour or per-month commit, billed
      directly or via cloud marketplaces.
    entries:
      - label: Hosted GPU instance
        name: dgx_instance
        type: metered
        metric: instance-hour
        limit: -1
        timeFrame: usage
        geo: configurable
        unit: 1
        price: contact sales
        userMultiplied: false
    elements:
      - name: Multi-cloud GPU access
      - name: Pre-configured AI training stack
      - name: Per-instance-hour or commit pricing
maintainers:
  - FN: API Evangelist
    email: info@apievangelist.com
x-canonical:
  generated: '2026-10-08'
  vocabulary: all/0-working/economics/plans-canonical.yml
  models:
  - free
  - hybrid
  - metered
  - subscription
  units:
  - compute
  - credit
  - negotiated
  - object
  plans:
  - name: build.nvidia.com Free Credits
    access: freemium
    model: free
    model_how: derived
    units:
    - credit
    periods:
    - per-use
  - name: NGC Catalog (Development)
    access: freemium
    model: metered
    model_how: derived
    units:
    - object
    periods:
    - per-use
  - name: NVIDIA AI Enterprise — 90-Day Trial
    access: freemium
    model: subscription
    model_how: derived
    units:
    - negotiated
    periods:
    - per-use
  - name: NVIDIA AI Enterprise
    access: enterprise
    model: hybrid
    model_how: derived
    units:
    - compute
    - other
    periods:
    - per-use
    - year
  - name: DGX Cloud / DGX Cloud Lepton
    access: other
    model: metered
    model_how: stated
    units:
    - compute
    periods:
    - per-use