Path forwards after PyTorch 2.6 Inference containers support ending June 30, 2026

[humanzz]

Hello team,

I had previously created the feature request [feature-request] Release SageMaker Inference containers for PyTorch 2.7 and 2.8 and the update there was that new inference versions were not keeping up with Torch (and training container) releases because of https://github.com/pytorch/serve going into maintenance mode.

I noted the update in https://aws.github.io/deep-learning-containers/#support-updates

• [2026/02/10] Extended support for PyTorch 2.6 Inference containers until June 30, 2026
• PyTorch 2.6 Inference images will continue to receive security patches and updates through end of June 2026

but it did not provide any update about the path forwards after that.

Are there plans for releasing new inference images, with versions >2.6?

[mabunday]

Hi Ahmed,

I'm working on a solution for this that will be launching in coming weeks - modern, secure training + inference containers with Python 3.10-3.14 (including free-threaded), revived multi-model server, and script-mode with any version of PyTorch/TensorFlow/SciKit-Learn/XGBoost (user chooses) in a single tag.

If you're interested please feel free to find me on LinkedIn - I would love to understand your current and future use-cases more.

Thanks,

Mark

[zhuofuAMZ]

Hi Ahmed,
Thank you for following up on this and for referencing your previous feature request.

Yes, the AWS DLC team is actively working on a new inference container solution to support the latest PyTorch versions, including versions beyond 2.6. This effort is in response to the maintenance mode status of PyTorch Serve.

While the project is still undergoing development and I cannot commit to a specific timeline at this point, I can confirm that the new solution will be available before the PyTorch 2.6 end of support date.