With vLLM via pip, run any of the models in the Llama 4 family with a simple command.

🙏 DeepSeek's highly performant inference engine is built on top of vLLM. Now they are open-sourcing the engine the right way: instead of a separate repo, they are bringing changes to the open source community so everyone can immediately benefit! github.com/deepseek-ai/op…

vLLM has just reached 50K github stars! Huge thanks to the community!🚀 Together let's bring easy, fast, and cheap LLM serving for everyone✌🏻

Thanks for the great write-up! 🙌 Prefix caching is critical for agentic workflows like @ManusAI_HQ , and vLLM makes it seamless. ✅ prefix caching is enabled by default with an efficient implementation ✅ Append-only context? Cache hit heaven Context engineering FTW 🚀

new post. there's a lot in it. i suggest you check it out

Why add sensors and complex systems when physics can do the job? This production line sorts products using only weight and controlled bursts of air. ✅ No cameras or vision models ✅ No expensive integration ✅ Just reliable, repeatable separation at scale It’s a reminder that

being old, senile, and online must feel something like a never-ending dream

Naval Ravikant revealed what most people never realize about the meaning of life:

How do you build and contribute to vLLM? In our latest vLLM Office Hours, we walked through the full contributor workflow from building vLLM across different hardware targets to submitting your first PR. Video: youtube.com/live/RjTuvQwob… Slides: docs.google.com/presentation/d…

PewDiePie in 2025: – built a 10×4090 rig – runs Llama 70B, gpt-oss-120B & Qwen 245B locally via vLLM – built a custom web UI (chat, RAG, search, TTS) – ran protein-folding simulations for charity – created an AI “council”, a swarm of 64 models – now fine-tuning his own model

Introducing OlmoEarth 🌍, state-of-the-art AI foundation models paired with ready-to-use open infrastructure to turn Earth data into clear, up-to-date insights within hours—not years.

My personal take: Some teams push the limits with massive scaling. Others focus on optimization under tight constraints. Both are needed. The trick is to stay honest about which game you’re actually playing..

AI’s next frontier is Spatial Intelligence, a technology that will turn seeing into reasoning, perception into action, and imagination into creation. But what is it? Why does it matter? How do we build it? And how can we use it? Today, I want to share with you my thoughts on

Humans are like CPUs, AI is like a GPU. The next bottlenecks aren’t compute, but bandwidth and verification: how fast we can talk to AI, and how well we can check its answers for correctness and safety.

Thanks to GitHub for spotlighting vLLM in the Octoverse 2025 report — one of the fastest-growing open-source AI projects this year. 🏆 Top OSS by contributors 🚀 Fastest-growing by contributors 🌱 Attracting the most first-time contributors Trusted by leading open model

🚀 No More Train–Inference Mismatch! We demonstrate bitwise consistent on-policy RL with TorchTitan (training) + vLLM (inference) — the first open-source run where training and inference numerics match exactly. It only takes 3 steps: 1️⃣ Make vLLM batch-invariant (same seq →

Introducing Marble by World Labs: a foundation for a spatially intelligent future. Create your world at marble.worldlabs.ai

In complex systems, scale doesn’t just add more; it rewires everything. At some threshold, the structure that was stable at small scale collapses and a new one takes over. That jump is what we casually call “from quantity to quality.”

Check out a great framework built upon vLLM, to serve Any-to-Any multimodal models!

🎉vLLM v0.11.2 is out! This release focuses on things the community cares about most — smoother scaling, more predictable performance, and wider model support. 1456 commits from 449 contributors (184 new!) made this possible. 🧡 Here are a few improvements you'll feel in real