generated: '2026-06-20'
method: searched
source: >-
  https://docs.nvidia.com/nim/large-language-models/latest/release-notes.html;
  https://status.nvidia.com; openapi/*-openapi.yml
versioning:
  scheme: semver
  container_current: 1.15.0
  api_surface: >-
    The OpenAI-compatible REST surface is unversioned in the path (/v1/...); the version
    that moves is the NIM container image tag (semver, e.g. meta/llama-3.1-70b-instruct:1.15.0).
  production_branch: >-
    NVIDIA ships "PB" (Production Branch) releases alongside feature releases for long-lived,
    stability-focused deployments (e.g. 1.14.0-pb5.1).
  docs: https://docs.nvidia.com/nim/large-language-models/latest/release-notes.html
deprecation:
  policy_url: https://docs.nvidia.com/nim/large-language-models/latest/release-notes.html
  sunset_header: false
  deprecation_header: false
  notes: >-
    Model and container lifecycle is communicated through release notes and the model
    catalog rather than RFC 8594 Sunset/Deprecation response headers. The legacy TRT-LLM
    backend was moved to legacy status in 1.15.0 (PyTorch backend became default).
sla:
  url: https://www.nvidia.com/en-us/data-center/products/ai-enterprise/
  notes: >-
    Production SLAs and long-term support are delivered through NVIDIA AI Enterprise;
    the free hosted tier at build.nvidia.com carries no uptime SLA.
status_page: https://status.nvidia.com
deprecated_operations: []