One stop shop for running AI/ML on AWS
Docs · Available Images · Tutorials
AWS Deep Learning Containers (DLCs) are pre-built Docker images for running AI/ML workloads on AWS. Each image is tested and patched for security vulnerabilities. For more details, visit our documentation.
- [2026/08/03] vLLM Server v2.2 (AL2023) — EC2:
server-cuda-v2.2· SageMaker:server-sagemaker-cuda-v2.2· vLLM0.26.0(up from 0.24.0); FlashInfer 0.6.15.post1; DeepEP EPv2/GIN backend (NCCL pinned to 2.30.7); Inkling (piecewise CUDA graph, MTP speculative decoding, LoRA, NVFP4), Cosmos3 Edge Reasoner, TranslateGemma-12b-it, BertForMaskedLM. - [2026/07/31] Base cu132 (CUDA 13.2, AL2023) — EC2:
devel-cu132-amzn2023·runtime-cu132-amzn2023· CUDA 13.2.1 with Python 3.13.12 (built from source) anduvpre-installed;develandruntimevariants;develbundles the multi-node stack (GDRCopy, NCCL 2.29.7, EFA installer). - [2026/07/26] vLLM v0.26.0 (Ubuntu) — EC2:
0.26.0-gpu-py312-ec2· SageMaker:0.26.0-gpu-py312· Inkling (piecewise CUDA graph, MTP speculative decoding, LoRA, NVFP4), Cosmos3 Edge Reasoner, TranslateGemma-12b-it, BertForMaskedLM; DeepSeek-V4 routing-kernel andfused_topk_biasspeedups; fp32lm_headviahead_dtype; per-KV-cache-group attention backends. - [2026/07/26] SGLang v0.5.16 (Ubuntu) — EC2:
0.5.16-gpu-py312-ec2· SageMaker:0.5.16-gpu-py312· Inkling day-0 support, LongCat 2.0 FP8, JetBrains Mellum v2, Pi0.5; DSpark speculative decoding (--speculative-algorithm DSPARK); UnifiedRadixTree now the default for SWA, Mamba, and DSA models. - [2026/07/20] PyTorch v2.13.0 — EC2:
2.13-cu133-amzn2023· SageMaker:2.13-cu133-amzn2023-sagemaker· Amazon Linux 2023 with EFA, flash-attn, and Transformer Engine; PyTorch 2.13.0 with CUDA 13.3.0, NCCL 2.30.7, TE 2.17.0, DeepSpeed 0.19.2. - [2026/07/15] vLLM v0.25.1 (Ubuntu) — EC2:
0.25.1-gpu-py312-ec2· SageMaker:0.25.1-gpu-py312· Patch release: defer TorchCodec FFmpeg import error to runtime (unblocks startup without system FFmpeg); guard mixed-dtype allreduce RMSNorm quant fusions (fixes NVFP4 garbage output). - [2026/07/13] vLLM v0.25.0 (Ubuntu) — EC2:
0.25.0-gpu-py312-ec2· SageMaker:0.25.0-gpu-py312· LLaVA-OneVision-2, Unlimited OCR, MOSS-Transcribe-Diarize, openai/privacy-filter, Hy3. - [2026/07/11] SGLang v0.5.15 (Ubuntu) — EC2:
0.5.15-gpu-py312-ec2· SageMaker:0.5.15-gpu-py312· GLM 5.2 Tuned, Hy3, HRM-Text, LocateAnything-3B. - [2026/07/10] TensorFlow v2.21.0 (SageMaker training) — SageMaker CPU:
2.21.0-cpu-py312-amzn2023-sagemaker· SageMaker GPU:2.21.0-gpu-py312-cu129-amzn2023-sagemaker· Amazon Linux 2023 with Python 3.12; GPU images ship CUDA 12.9.1.
- [2026/04/28] We cannot guarantee security patching on Ubuntu-based vLLM and SGLang images due to the lack of Ubuntu Pro licensing. Customers may continue using these images at their own discretion and risk. We recommend migrating to our Amazon Linux-based images.
- [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
- For complete framework support timelines, see our Support Policy
- Distributed Training on Amazon EKS - Configure and validate a distributed training cluster with DLCs on Amazon EKS.
- DLCs with Amazon SageMaker AI & MLflow - Use DLCs with SageMaker AI managed MLflow for experiment tracking and model management.
- LLM Serving on Amazon EKS with vLLM - Deploy and serve LLMs on Amazon EKS using vLLM DLCs.
- Fine-tuning Meta Llama 3.2 Vision - Fine-tune and deploy Llama 3.2 Vision for web automation using DLCs, Amazon EKS, and Amazon Bedrock.
- DLCs with Amazon Q Developer and MCP - Streamline deep learning environments with Amazon Q Developer and Model Context Protocol.
- LLM Deployment on Amazon EKS - Deploy and optimize LLMs on Amazon EKS using vLLM DLCs. See also: Sample Code
This project is licensed under the Apache-2.0 License.
