Woosuk Kwon

@woosuk_k

@inferact | @vllm_project | prev: PhD @Berkeley_EECS

woosuk.me

Joined April 2023

848 Following

9,313 Followers

Replies Reposts Media

Pinned
Woosuk Kwon
@woosuk_k
Jan 22
Today, we're proud to announce @inferact, a startup founded by creators and core maintainers of @vllm_project, the most popular open-source LLM inference engine. Our mission is to grow vLLM as the world's AI inference engine and accelerate AI progress by making inference cheaper
183
Woosuk Kwon
@woosuk_k
15h
vLLM is all you need to reproduce DeepSeek's performance.
vLLM
@vllm_project
15h
🚨🚨🚨 vLLM is all you need to reproduce DeepSeek’s performance. With Wide-EP and EPLB, vLLM can match DeepInfer’s decode performance within 6% and generate $39/gpu/hr in revenue at DeepSeek’s off-peak pricing, at 200 P90 TPS on the AgentX workload. Specifically: • Under
1
Woosuk Kwon
@woosuk_k
Oct 20
Amazing release! 🚀 DeepSeek was an early adopter and contributor to vLLM, and they’ve kept giving back to the community, sharing their learnings through various releases. This is the magic of open source: building together and pushing the frontier forward! 🔥
DeepSeek
@deepseek_ai
Oct 20
🚀 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
3
Woosuk Kwon
@woosuk_k
Oct 16
Congrats to @reflection_ai on the release! This time, we let AI agents drive model optimization autonomously, with performance gains tracked at every step. So glad to see this working so well.
vLLM
@vllm_project
Oct 16
01:00
🎉 Reflection-Beam-1.0 from @reflection_ai has Day-0 support in vLLM. 501B total parameters, 23B active per token, and a 1M-token context window. Beam combines hybrid attention with Harmony-format reasoning and tool use. Beam also has Day-0 RL support in SkyRL (@NovaSkyAI):
7
Woosuk Kwon
@woosuk_k
Oct 16
@vllm_project powered the RL inference of Muse. Virtuous cycle of open source and open models. 🫡
Zhuohan Li
@zhuohan123
Oct 16
Excited to see Muse Spark Open released! 🚀 Proud that vLLM powered RL inference during its training at Meta. A glimpse of recursive self-improvement: RL makes models better; LLMs help us optimize inference; faster inference enables more RL. The flywheel keeps spinning. 🔁
Quote
vLLM
@vllm_project
Oct 15
🎉 Day-0 support for Meta's open-weight Muse Spark Open is live on vLLM, verified on NVIDIA and AMD GPUs as well as on TPUs! 816B parameters, 42B active per token, a 1M-token image, video and audio inputs, built for long-horizon agentic coding. We’re excited to have worked
2

Sunbelt Computer Software

PL/B Language Development and Support

Woosuk Kwon