🚀Excited to see our collaboration with LMSYS Org bring Multiple Token Prediction (MTP) in SGLang to production! Proud to support faster, smarter open-source LLM serving. #EigenAl #MTP #SGLang #LLMinfra #ModelServing #DeepSeek #OpenSourceAl #AskChatGPT

SGLang is now officially supporting OpenAI’s new GPT-OSS model!

Proud to collaborate with the SGLang team, it was a lot of fun!

Check out our blog post on the architecture deep-dive of GPT-OSS → blog.eigenai.com/2025-08-09-day… 5 days ago → we pulled off Day 0 serving for OpenAI’s GPT-OSS-120B & 20B on Hopper 🖥️ + Blackwell ⚡ — fast, stable, production-ready the moment the weights dropped. 🌀 Tuned Attention

Reminder: SGLang x AMD SF Meetup is this Friday (Aug 22) at Shack15 🎉 Hands-on GPU workshop, talks from AMD/xAI/SGLang, food + networking . Don’t miss it! 📣 Seats are filling fast - if you haven’t registered yet, now’s the time. Bring a friend! 👉 Register here:

SGLang LMSYS Org now delivers major performance boosts for gpt-oss-120b: 🚀 • 2.1× higher prefill throughput • 2.25× higher decode throughput • Optimized for Hopper, Blackwell & MI350 GPUs Try it out on our GPT-OSS Playground! 👉 chat.eigenai.com See the detailed