LMSYS Org Launches SGLang Consortium
- •LMSYS Org launched the SGLang Consortium with eight founding members at the inaugural SGLang Summit in San Francisco.
- •Deloitte estimates inference will account for roughly two-thirds of AI compute in 2026, compared with about half in 2025.
- •Members will hold two roadmap meetings yearly; SGLang maintainers retain technical decisions and code acceptance.
LMSYS Org, the nonprofit behind the open-source AI inference engine SGLang, launched the SGLang Consortium on November 12, 2026, with NVIDIA, AMD, Intel, Google, Meta, Cloudflare, Modal and RadixArk as founding members. Announced in the opening keynote of the inaugural SGLang Summit in San Francisco, the consortium will work with SGLang maintainers on the project’s roadmap and open-source AI inference.
Deloitte estimates inference will account for roughly two-thirds of all AI compute in 2026, up from about half in 2025. SGLang is a full-stack inference engine for frontier AI within the PyTorch ecosystem. It runs on hundreds of thousands of GPUs and powers large production deployments for Meta, Microsoft, Google, Amazon, Thinking Machines Lab, Cloudflare, Reflection and Modal.
The consortium will provide a standing forum for members and maintainers to discuss inference trends and coordinate engineering work on frontier model support, reasoning and multimodal workloads, distributed inference and hardware optimization. Each founding member will designate a technical representative and attend two roadmap meetings a year; contributions may include engineering expertise, deployment feedback, testing resources or joint development. Working groups may work between meetings, with public updates on roadmap direction and progress. SGLang maintainers retain technical decision-making and code acceptance, and the project remains open to the full community.
Ying Sheng, co-author of the original SGLang paper and RadixArk co-founder and CEO, said software approaches to AI efficiency would be important amid challenges in building data centers, and that lessons from large-scale SGLang deployments would benefit users and the AI ecosystem. NVIDIA vice president Ian Buck said better software helps developers get more performance from NVIDIA compute, open models support new applications and businesses, and lower inference costs make AI more accessible. The first SGLang Summit ran November 12–13 at San Francisco’s Fort Mason Center for Arts & Culture, with more than 100 speakers and sessions on reinforcement learning, multimodal models and hardware platforms.
The fastest, most reliable AI news, straight to your inbox
