Today at PyTorch Conference 2026, we announced llm-d 1.0! 🎉 Distributed inference is too complex to solve behind closed doors. 17 months in, llm-d is now a @CloudNativeFdn project powering fleets worldwide. Here’s what’s inside 1.0 🧵👇

🚀 Introducing DeepInfer, our in-house inference framework, now fully open source! Powering DeepSeek-V4.1-Flash at scale with high throughput and low latency. We’re sharing our production stack to help the community lower compute costs, reduce energy use, and make AI more affordable. 🔗 github.com/deepseek-ai/DeepI… 1/4

We still want to find a way to bring the community together. If you planned to be in California for the Grace Hopper Celebration, I invite you to join us this year for GitHub Universe, Oct 28-29. 🌉 We've set aside 75 complimentary passes for folks with cancelled registrations. While this doesn't replace the conference and Universe has far fewer attendees than GHC, my hope is that it does give folks an opportunity to connect with each other. 📷 DM the GitHub team on LinkedIn or email me at [email protected] (first come first serve).

An IBM slide from 1979.

Beam weights are out today (Apache 2.0 license). We integrated Beam across 40+ different partners. And are also open sourcing Mirror CLI - the harness Beam was trained against. Happy building.

Today we’re releasing Muse Spark Open, the most capable US open-weight model to date. Built for coding and agentic work, it excels at long-horizon tasks, complex software engineering, and multi-step reasoning. Key Model Specs: → 816B total / 42B active parameters → 1M token context window → Natively multimodal across text, image, video, audio

🎶 CALLING ALL MY OLD SCHOOL HIP HOP HEADS 🎶 DE LA SOUL is playing at the AI Community Bash the day before AGNTCon + MCPCon in San Jose! We've hidden some of their song titles in this blog post. Find one and I have free ticket for you! 🕵🏻‍♂️ aaif.io/blog/agent-identity-…