First SGLang Summit draws 2,000+; v1.0 ships with 9-member consortium incl. NVIDIA, Google
BanghuaZ · x · 2026-11-18
The first SGLang Summit ran for two days with 2,000+ attendees, 90+ technical talks and six workshops. Key announcements:
- SGLang v1.0 released, plus Miles v0.2
- The SGLang Consortium launched with nine founding members: NVIDIA, AMD, Intel, Google, Meta, Thinking Machines, Cloudflare, Modal, and RadixArk
- SGLang is now used by 332 organizations; at Meta it powers production inference for products serving billions of users
- Production lessons shared by Modal, Cloudflare, LinkedIn and others
- One talk covered multi-day agent-written kernel campaigns, bridging "it runs" and "it's fast"
Themes spanned the future of inference, RL, and AI agents.
More from coding & agent
- Grok-powered agent given $50 and a 'survive or shut down' challenge grew it to $5,763 in 48 hours — nikola_mr64990 · 2026-11-18
- Sable: an open-source desktop agentic dev environment wrapping Claude Code and Codex CLIs — churchkey · 2026-11-18
- Qwen3 4B scores 70/99 in a bare agent loop but 55/99 in LangChain Deep Agents with 4x tokens — SeveralSeat2176 · 2026-11-18
- OpenHands Teams launches: self-service agents to automate CVE remediation — rajistics · 2026-11-18
- Agent task trees deadlock under concurrency caps — slot lending fixes it — arpit_bhayani · 2026-11-18
- Developer builds a sandbox world where you can watch your AI agents interact — Agent_City10 · 2026-11-18