apiVersion: catalog.confighub.com/v1alpha1
kind: NIMModelProfileRecord
metadata:
  name: llama-3-1-8b-instruct-1xgpu
spec:
  description: >-
    The first described model profile of the inference entry: the smallest
    current-generation shape in the retained upstream matrix. This record is
    configuration data about the shape; it grants nothing and fetches nothing.
  modelShape: llama-3-1-8b-instruct-1xgpu
  modelShapeFile: upstream/kserve/nim-models/llama-3.1-8b-instruct_1xgpu_1.1.0.yaml
  servingRuntime: nvidia-nim-llama-3.1-8b-instruct-1.1.0
  servingRuntimeFile: upstream/kserve/runtimes/llama-3.1-8b-instruct-1.1.0.yaml
  image: nvcr.io/nim/meta/llama-3.1-8b-instruct:1.1.0
  gpuCount: 1
  storageUri: pvc://nvidia-nim-pvc/
  licensing:
    imageRegistry: >-
      nvcr.io is NGC-gated. The image is pulled only by the user's cluster with
      the user's NGC API key under the user's own NVIDIA entitlement.
    ngcCatalogPage: https://catalog.ngc.nvidia.com/orgs/nim/teams/meta/containers/llama-3.1-8b-instruct
    governingTermsReadAt: "2026-08-07"
    governingTermsNamed:
      - NVIDIA Software License Agreement
      - Product-Specific Terms for AI Products
      - NVIDIA Open Model Agreement
      - Llama 3.1 Community License Agreement
    governingTerms: >-
      The four names above were read from the NGC catalog page on 2026-08-07.
      Re-read the per-artifact governing terms at deploy time; they override
      any general statement here. The model carries its own model license in
      addition to NVIDIA's product terms.
    licenseRead: docs/planning/nim-ngc-license-read.md
status:
  result: described-offline
  imagePulled: false
  modelFetched: false