Your open-source AI ally. We are committed to making production-grade AI inference as accessible and reliable as electricity, powered by vLLM.
Joined October 2023
- Tweets625
- Following1.4K
- Followers1.3K
- Likes1K
@AMD MI350X Server just landed at Embedded LLM.
Deeply grateful to for @AIatAMD this generous gift and continued support.
More compute for the @vllm_project ROCm work we’ve been leading. More results like our AgentX optimization work with AMD.
More to give back to the community.
We helped close the performance gap.
@SemiAnalysis_ Kimi K3 estimator puts MI355X at 36% higher revenue per chip-hour than HGX B300, with a 38.8% profit margin vs 7.7%.
Performance = profit.
🚀 @AMD MI355X just showed competitive performance to NVIDIA B300 on @Kimi_Moonshot K3 serving!
8× MI355X delivers 24.4% more throughput than HGX B300 (7,567 vs 6,083 tok/s/GPU) at 50 tok/s/user on @SemiAnalysis_ AgentX agentic coding benchmark, and 87.4% more tokens per dollar.
We cover the vLLM optimizations, AMD kernels, and hardware details that closed the gap, plus patterns for spotting bottlenecks so you can get more out of vLLM on your own GPUs.
Thanks @AMD @embeddedLLM @SemiAnalysis_
Full writeup: 🔗 vllm.ai/blog/2026-10-16-kimi…
Meet us at #COLM2026 in San Francisco.
Oct 9 at DAIH: our research on deploying clinical AI where the data lives across Nvidia and AMD GPU, with energy and cost in view.
Building models or putting them into production? Come say Hi! @manpree59181175
DM us to meet.
Massive respect to @AMD, @roaner and @AnushElangovan for investing where it counts: dedicated hardware, CI, and the room for core maintainers @Rxday000 to push ROCm to its limit.
The partnership is in a better place than it has ever been. 502 commits is the baseline. When software and silicon move at the same speed, the whole open source ecosystem wins. Proud of what we’re building together on @vllm_project.
#1 company contributing code to @vllm_project right now?
@AMD: 502 commits to core vLLM in 90 days. 13% of all org contributions, almost 4x Nvidia.
Also grateful to Red Hat, IBM, Embedded LLM & Inferact for building it with us.
Open source wins when hardware has a choice.
We squeezed every last microsecond out of the kernels, then made the agent resend 200K tokens just to say “continue.”
Yeah, that hurts.
"vllm-project/agentic-api": stateful continuation, smaller requests, lower latency, Codex + Claude Code + etc, MCP, server tools, and web search for vLLM.
Your harness will thank you.
X called our colleague Mei Ling "your competition."
Open-source AI isn’t a competition. It’s a collaboration.
Here she is with the Embedded LLM team as we partner with TAIONE to grow Taiwan’s vLLM ecosystem:
taione.org/news/vllm-embedde…
Link to her original post: lnkd.in/p/gHP-FraW
Our Tun Jian Tan @Rxday000 will be speaking at the first vLLM Conference (co-located with Ray Summit 2026) next week in San Francisco.
Talk: agentic-api - A Stateful Agent Layer for vLLM
Most teams still can’t run Claude Code or Codex the native way against their own @vllm_project. Multi-turn state, WebSockets, tool loops, and server-side web search are missing.
That’s the layer we’ve been building with @franciscojarceo @RedHat_AI.
Tun will show how vllm-project/agentic-api sits in front of vLLM and delivers the full native agentic experience, and why this is the layer cloud API providers will use to offer first-class agent APIs.
If you’re building real agents on open models, this is one of the talks you don’t want to miss.
Aug 26 · 2pm San Francisco Marriott Marquis
Zero Day @vllm_project support on AMD MI455X / Helios
Proud of our collaboration with @AMD and @inferact
🚀 Speed is the MOAT! The ecosystem that ship faster win.
Official ROCm nightly wheels + Docker images are now live for vLLM!
Seamless daily updates with ROCm 7.2.1, PyTorch 2.10 & Triton 3.6.
Thanks for raising this @SemiAnalysis_ big win for the ecosystem.
Props to @AIatAMD and @vllm_project team! Awesome working with you all! 🔥
Making ROCm a first-class citizen in vLLM. ✅
This is what happens when Embedded LLM and @AMD join forces. 🤝
Native ROCm 7.0 support is now front and center on the official @vllm_project Quick Start guide. We are proud to have partnered with the @AIatAMD team to make high-performance inference this accessible.
Select ROCm ➡️ Copy ➡️ Run. It’s now that simple. 🚀
