Helion, PyTorch's domain-specific language (DSL) for custom kernels, is expanding toward hardware-heterogeneous kernel authoring. Read the full technical deep dive here 👉 https://bit.ly/4fE6qRs..*

As the officially recommended framework for the AMD AI DevMaster Hackathon, PyTorch enables developers to build AI applications on ROCm-enabled AMD GPUs. Choose a Multimodal AI, Agentic AI, or Physical AI track. Submissions close Aug. 6: https://luma.com/amd-4dhi

The PyTorch Conference North America schedule is live. The Oct. 20–21 event in San Jose spans training, inference, compilers, responsible AI, applications, and PyTorch Foundation projects. The Poster CFP closes July 26. https://pytorch.org/blog/pytorch-conference-north-america-schedule-is-live/

TODAY: Matt White, PyTorch Foundation CTO, will discuss right-sized models, routing, caching, and task-level measurement at AMD Advancing AI on July 22 at 11:50 a.m. PDT. https://bit.ly/4bXUNDB #AdvancingAI

We are excited to launch a new quarterly blog series where each hosted project - PyTorch, vLLM, DeepSpeed, Ray, Helion and Safetensors - shares its latest updates, technical progress, and roadmap. Read the complete update to learn more 👉 https://bit.ly/4fkiTJy..*

Tomorrow, July 22, at 11 a.m. PT, Alban Desmaison, Andrey Talman, Piotr Bialecki, and moderator Chris Gottbrath will cover PyTorch 2.13, including FlexAttention, CuTeDSL, and FSDP2 updates, and answer questions live. Register: https://pytorch.org/event/pytorch-2-13-release-live-qa/

PyTorch maintainers and contributors will provide a brief overview of PyTorch 2.13 and answer questions live on July 22 at 11 a.m. PT. Panelists: Alban Desmaison, Andrey Talman, and Piotr Bialecki, moderated by Chris Gottbrath. Register: https://pytorch.org/event/pytorch-2-13-release-live-qa/

The PyTorch Certified Associate program is designed to help early-stage practitioners build instant credibility and validate their real-world skills in AI and machine learning. Learn more about the program and how to enroll here: https://bit.ly/4yp6jlb..*🔥

This week at WAIC 2026 in Shanghai, PyTorch Foundation Executive Director Mark Collier and CTO Matt White will speak July 18–19 at Alibaba Cloud, BAAI, and Ant Group forums on Agentic Cloud, open source AI infrastructure, and accessible AGI. 📸 Mark spoke with CCTV.

The Triton Plugin Extensions system is a framework for dynamically loading custom compiler passes, dialect ops, and DSL extensions into upstream Triton at runtime, without forking or recompiling. Read the technical deep dive and how to get started here 👉 https://bit.ly/4fATzAu

PyTorch Foundation Contributor Awards celebrate outstanding impact across PyTorch Foundation hosted projects, including PyTorch, vLLM, DeepSpeed, Ray, Helion, and Safetensors🔥Submit your nomination by July 17 👉 https://bit.ly/4vnMKa3

PyTorch 2.13 is here: 3,328 commits from 526 contributors, with updates across FlexAttention, CuTeDSL, nn.LinearCrossEntropyLoss, torchcomms, FSDP2, Python 3.15 wheels, ROCm, Arm, and XPU. Read the blog and register for the July 22 live Q&A: https://pytorch.org/blog/pytorch-2-13-release-blog/

AMD and Meta contributors detail the ROCm port that brings PyTorch Monarch to AMD Instinct GPUs. Validation includes Llama 3 8B on 128 GPU MI300 and 256 GPU MI355 clusters. https://pytorch.org/blog/bringing-pytorch-monarch-to-amd-gpus-single-controller-distributed-training-on-rocm/

Ever wondered why a PyTorch CI test failure name doesn't exactly match your source file? Read our latest blog which provides a contributor's perspective on how to get started with testing in PyTorch. Link in comments 👇

PyTorch Foundation supported the ExecuTorch Hackathon, where teams built on-device AI apps with PyTorch and ExecuTorch. Congratulations to SafeScreen AI, SixthSense and Toddle AI, the winners. Read recap: https://pytorch.org/blog/building-the-future-of-on-device-ai-at-the-executorch-hackathon/

PyTorch Foundation is a Gold Sponsor of Agentic AI Summit 2026. Matt White, CTO of PyTorch Foundation, will lead “The Open Agentic Stack,” on building AI systems with open source, open standards, and composability. 🔗 https://pytorch.org/event/agenticai-summit/

We’re excited to welcome Shopify as a Platinum Member of the PyTorch Foundation 🎉 Shopify provides essential internet infrastructure for commerce and relies on PyTorch as a key part of applied machine learning across Shopify's all-in-one platform that makes it easier to start, run, & grow a business

Built on PyTorch, Ray, SGLang, and NVIDIA Megatron-LM, RadixArk’s Miles is an open source framework for large-scale LLM RL post-training with async rollout/training, MoE alignment, LoRA, and extensibility. 🔗 https://pytorch.org/blog/miles-a-pytorch-native-stack-for-large-scale-llm-rl-post-training/

PyTorch now has a Cross-Repository CI Relay (CRCR) that streamlines testing across the ecosystem. CRCR automatically triggers and tracks continuous integration in out-of-tree backend repositories whenever a PR is opened against pytorch/pytorch. Link to learn more in comments section 👇

One runtime, multiple GPU architectures, and zero vendor-specific model code. In this blog post, the TokenSpeed team at Lightseek introduces TokenSpeed-Kernel, a portable, high-performance kernel system built for modern LLM inference. Link to blog in comments section 👇

When an AI agent recognizes an action is unsafe but executes it anyway, Besmira Nushi of NVIDIA calls it an “action-thought disconnect.” In this #PyTorchCon keynote, she explains why trajectory evaluation can reveal failures task scores miss. https://youtu.be/54u05pEJv6g?si=VwS0m5Dd5X98mX_S

While SGLang provided Day-0 support for DeepSeek-V4, the collaboration between the SGLang and NVIDIA engineering teams has taken its production performance to the next level. Find the full technical breakdown in the comments below:

Currently Helion searches utilize Likelihood-Free Bayesian Optimization (LFBO) to find the most performant configs. In this blog, we look at how LLM-guided autotuning is a practical approach to dramatically faster kernel tuning at production quality. Click the link in the comments to learn more.

Prove you can work with the technology behind modern AI: We are excited to announce the launch of PyTorch Certified Associate (PTCA). Link to learn more in comments section👇

The schedule is now available for KubeCon + CloudNativeCon + OpenInfra Summit + PyTorch Conference China, September 7-9 in Shanghai. AI is transforming how we build, deploy, and operate technology. Open source is making it possible. Explore the agenda: https://bit.ly/4ezq2W9

Nominations are officially open for the 2026 PyTorch Foundation Contributor Awards. These awards celebrate outstanding impact across PyTorch Foundation hosted projects, including PyTorch, vLLM, DeepSpeed, Ray, Helion, and Safetensors. Submit your nomination by July 17. Link in comments

We're seeking a Senior Cloud Operations Engineer who will play a pivotal role in the PyTorch Foundation, leading cloud infrastructure and DevOps initiatives. Apply at: careers.linuxfoundation.org/jobs/647914-...

24 hours left to nominate yourself or someone else as a PyTorch Foundation Ambassador🔥 We especially welcome applications from contributors in Africa, Latin America, the Middle East, Oceania, Southeast Asia, & Eastern Europe. Learn more and apply before June 18, 2026 - link in comments.

The inaugural PyTorch Meetup Singapore brought together engineers, researchers, and community builders to talk about everything from the vLLM project to sovereign intelligence. Read the full technical recap and find presentation slides in our latest blog: https://bit.ly/4vdcPJU

Helion brings PyTorch-native kernel development to vLLM. Xiaohong Chen (Red Hat) and Yanan Cao (PyTorch, Meta) integrated Helion kernels into vLLM for FP8 inference with Qwen3 models, evaluating fused GPU kernels across H100 and B200 GPUs. Read more: https://bit.ly/4xl72TK