Announcing AWS Neuron SDK 2.26.0
Today, AWS announces the general availability of Neuron SDK 2.26.0, delivering improvements for deep learning workloads on AWS Inferentia and Trainium-based instances. This release introduces support for PyTorch 2.8 and JAX 0.6.2, along with enhanced inference capabilities on Trainium2 (Trn2) instances. These updates enable developers to leverage the latest frameworks while benefiting from improved model deployment flexibility and performance optimizations.
With Neuron SDK 2.26.0, customers can now deploy FLUX.1-dev image generation model, along with Llama 4 Scout and Maverick variants (beta) on Trn2 instances. The release introduces expert parallelism support (beta) for efficient distribution of Mixture-of-Experts (MoE) models across multiple NeuronCores, and adds new capabilities through new Neuron Kernel Interface (NKI) APIs. The updated Neuron Profiler provides improved capabilities, including system profile grouping for distributed workloads.
The new SDK version is available in all AWS Regions supporting Inferentia and Trainium instances, offering enhanced performance and monitoring capabilities for machine learning workloads.
To learn more and for a full list of new features and enhancements, see: