Is this page helpful?

End of Life Notices#

This page lists NVIDIA AI Enterprise components that have reached or are approaching Deprecated or End of Life. For definitions of lifecycle states (Supported, Deprecated, End of Support, End of Life), refer to Lifecycle State Definitions.

Table 23 Active Notices - Deprecated#

Component

Current State

Action Required By

Deprecated hardware (Infra 8.0+) - V100, RTX 4000 SFF Ada, RTX A4000, Quadro RTX 8000 / 6000 / 4000

Removed starting in 8.0 (still supported on Infra 7.x LTSB; Infra 4.10 LTSB reached EOL July 2026)

Migrate or remain on a supported LTSB branch

NVIDIA NIM Retrieval QA E5 Embedding v5

Deprecated

January 2027 (PB 6)

NVIDIA TensorRT-LLM Backend (*-py3-trtllm)

Deprecated (not included in PB6)

Before upgrading to Production Branch 6 (PB6)

Warning

The following components have reached End of Support with an Immediate action date. Do not start new deployments on these components. Follow the detailed migration sections below for each entry.

Table 24 Active Notices - Immediate End of Support#

Component

Current State

Action Required By

NVIDIA Morpheus

End of Support

Immediate

NVIDIA NIM Llama-3.1-70b-instruct

End of Support (Model Replaced)

Immediate

NVIDIA Optimized TensorFlow 2 Containers

End of Support

Immediate

Deprecated Hardware (Infra 8.0+)#

The following GPUs are no longer supported for vGPU for Compute starting in NVIDIA AI Enterprise Infra 8.0 (and remain unsupported on Infra 8.1). They remain supported on the Infra 7.x LTSB branch for the remainder of that LTSB support window. The Infra 4.10 LTSB branch reached end of life in July 2026.

Components

  • NVIDIA Tesla V100 (all Volta variants - V100 SXM2 16 GB / 32 GB, V100 PCIe 16 GB / 32 GB, V100S PCIe 32 GB, V100 FHHL)

  • NVIDIA RTX 4000 SFF Ada Generation (Ada Lovelace)

  • NVIDIA RTX A4000 (Ampere)

  • NVIDIA Quadro RTX 8000 (Turing)

  • NVIDIA Quadro RTX 6000 (Turing)

  • NVIDIA Quadro RTX 4000 (Turing)

Current State

Removed from supported GPUs starting in Infra 8.0; still supported on Infra 7.x LTSB. Infra 4.10 LTSB reached end of life in July 2026.

Deprecation Announced

May 2026 (Infra 8.1 GA, applied retroactively to Infra 8.0 as a hotfix)

Last Supported Production Branch Release

None on the 8.x line - these GPUs are removed from 8.0 onward.

First Unsupported Release

Infra 8.0

Continuing LTSB Support

Infra 7.x LTSB continues to support these GPUs through its LTSB support window. Infra 4.10 LTSB reached end of life in July 2026.

Timeline

  • November 2025 - Original Infra 8.0 GA shipped with V100 and the 5 workstation-class GPUs supported.

  • May 2026 - Infra 8.1 GA removes V100 (Volta) and the 5 workstation-class GPUs from the supported GPU list. The change is also applied retroactively to Infra 8.0 as a post-publish hotfix so the 8.x line is consistent.

  • July 2026 - Infra 4.10 LTSB reached end of life. These GPUs are no longer available through the 4.x branch.

  • Ongoing - Infra 7.x LTSB continues to support these GPUs for the remainder of its LTSB support window.

Migration Path

Customers running these GPUs have two options:

  1. Migrate to a currently supported GPU on Infra 8.x:

    • Turing - NVIDIA T4 (16 GB)

    • Ampere - NVIDIA A10, A16, A30, A40, A100 (40 GB / 80 GB), RTX A5000, RTX A6000

    • Ada Lovelace - NVIDIA L4, L20, L40, L40S, RTX 5000 / 5880 / 6000 Ada Generation

    • Hopper - NVIDIA H100, H200, H800, H20

    • Blackwell - NVIDIA B200, B300, RTX PRO 4500 / 6000 Blackwell

  2. Remain on a supported NVIDIA AI Enterprise Infra LTSB branch (Infra 7.x LTSB) where these GPUs continue to be supported through the LTSB support window. (The Infra 4.10 LTSB branch reached end of life in July 2026 and is no longer a supported option.)

Refer to the Infra 8.1 Release Notes “Deprecations and Removals” section and the Infra 8.1 Support Matrix for the complete list of supported GPUs on 8.x.

Action Required

Choose one of the two migration options above. The deprecated GPUs have been removed from the 8.x documentation set (Support Matrix, vGPU types reference, vGPU features tables, glossary, and Linux memory limitations) and from the Infrastructure Lifecycle and Compatibility Explorer for Infra 8.0 and 8.1.

NVIDIA NIM Retrieval QA E5 Embedding v5#

Component

NVIDIA NIM Retrieval QA E5 Embedding v5

Current State

Deprecated

Deprecation Announced

January 2027 (PB 6)

Current Version (PB6)

1.14.0

Timeline

  • October 2024 - E5 Embedding v5 version 1.2 included in Production Branch - October 2024 (PB 24h2).

  • May 2025 - E5 Embedding v5 version 1.8.0 included in Production Branch - May 2025 (PB 25h1).

  • October 2025 - E5 Embedding v5 continues in Production Branch - October 2025 (PB 25h2) alongside NVIDIA Retrieval QA Llama 3.2 1B Embedding v2.

  • January 2026 - E5 Embedding v5 version 1.11.2 released in PB 25h2.

  • May 2026 - E5 Embedding v5 continues in Production Branch 6 (PB6); version 1.14.0 released.

  • July 2026 - Production Branch - October 2025 (PB 25h2) reached End of Life.

  • January 2027 - Production Branch 6 reaches End of Life. E5 Embedding v5 reaches End of Support.

Migration Path

Transition to NVIDIA Retrieval QA Llama 3.2 1B Embedding v2, which is available alongside E5 Embedding v5 in the current Production Branch 6 (PB6).

Action Required

Plan migration to NVIDIA Retrieval QA Llama 3.2 1B Embedding v2 before January 2027 (PB 6). After deprecation, E5 Embedding v5 will no longer receive CVE patches, bug fixes, or updates.

NVIDIA TensorRT-LLM Backend#

Component

NVIDIA TensorRT-LLM backend container images (*-py3-trtllm)

Current State

Deprecated (not included in Production Branch 6)

Last Supported Production Branch Release

Production Branch 5.9 (PB 5.9)

First Unsupported Production Branch Release

Production Branch 6 (PB6)

Timeline

  • PB 5.9 - Final Production Branch release to include the TensorRT-LLM backend (*-py3-trtllm) Triton container variant.

  • May 2026 - Production Branch 6 (PB6) released without a TensorRT-LLM backend container variant; PB6.x monthly updates do not include *-py3-trtllm.

Migration Path

Transition to the vLLM backend Triton container variant (*-py3-vllm), which is included in Production Branch 6 (PB6). For NVIDIA NIM deployments, refer to the NVIDIA NIM for LLMs documentation for supported inference backends on the current Production Branch.

Action Required

Customers using the TensorRT-LLM backend on earlier Production Branch releases should plan migration to the vLLM backend before upgrading to Production Branch 6 (PB6). The *-py3-trtllm container variant is not available in PB6 or subsequent PB6.x monthly releases.

NVIDIA NIM Llama-3.1-70b-instruct#

Component

NVIDIA NIM Llama-3.1-70b-instruct inference microservice

Current State

End of Support (Model Replaced)

End of Support Date

July 2026

Last Supported Version

1.10

Timeline

  • October 2024 - Llama-3.1-70b-instruct NIM version 1.3 included in Production Branch - October 2024 (PB 24h2).

  • May 2025 - Llama-3.1-70b-instruct NIM version 1.10 included in Production Branch - May 2025 (PB 25h1).

  • October 2025 - Llama-3.1-70b-instruct included in Production Branch - October 2025 (PB 25h2).

  • July 2026 - Production Branch - October 2025 (PB 25h2) reached End of Life. Llama-3.1-70b-instruct reaches End of Support.

Migration Path

Transition to NVIDIA NIM Llama-3.3-70b-instruct, which is available in the latest Production Branch. Alternatively, use the Multi-LLM NIM for flexible model deployment.

Action Required

Customers still using NVIDIA NIM Llama-3.1-70b-instruct should migrate to NVIDIA NIM Llama-3.3-70b-instruct or a supported NIM model. No further updates will be provided for this model after July 2026.

NVIDIA Morpheus#

Component

NVIDIA Morpheus cybersecurity AI framework container images

Current State

End of Support

End of Support Date

January 2026

Last Supported Version

25.02-runtime

Timeline

  • May 2024 - Morpheus 24.02-runtime included in Production Branch - May 2024 (PB 24h1).

  • October 2024 - Morpheus 24.06-runtime included in Production Branch - October 2024 (PB 24h2).

  • May 2025 - Morpheus 25.02-runtime included in Production Branch - May 2025 (PB 25h1). This was the final Production Branch to include Morpheus.

  • October 2025 - Morpheus was not included in Production Branch - October 2025 (PB 25h2).

  • January 2026 - Production Branch - May 2025 (PB 25h1) reached End of Life. Morpheus is no longer supported within NVIDIA AI Enterprise.

Migration Path

Transition to the Morpheus GitHub repository, which is maintained on a best-effort basis. Community contributions are accepted, and CVE fixes are provided as possible. No new feature development is planned.

Action Required

Customers still using NVIDIA Morpheus containers should migrate to an alternative solution immediately. No further CVE patches, bug fixes, or updates will be provided.

NVIDIA Optimized TensorFlow 2 Containers#

Component

NVIDIA Optimized TensorFlow 2 container images

Current State

End of Support

End of Support Date

July 2025

Last Supported Version

24.08-tf2-py3

Timeline

  • February 2025 - TensorFlow 2 24.02 release was the final Feature Branch update.

  • June 2025 - October 2024 Production Branch (24h2) received its final monthly security update.

  • July 2025 - TensorFlow 2 containers reached End of Support and are no longer supported within NVIDIA AI Enterprise.

Migration Path

Transition to alternative solutions such as JAX with Keras. Refer to the NVIDIA JAX Toolbox for guidance on JAX and the NGC Catalog for other supported frameworks within NVIDIA AI Enterprise.

Action Required

Customers still using NVIDIA Optimized TensorFlow 2 containers should migrate to a supported framework immediately. No further CVE patches, bug fixes, or updates will be provided.