NEWS BRIEF
What happened
Cloudflare has published an incident account involving cross-request KV cache exposure in its GLM-5.3 deployment on Workers AI. The company describes a configuration mismatch combined with an SGLang rejection-path bug. It says it disabled the model, corrected the configuration, patched the inference engine, and restored service on the day of its response.
What the announcement does not establish
The account concerns a specific deployment and failure path. It does not establish that every GLM or SGLang deployment was affected, nor that unrelated inference services share the same flaw.
Primary source: Cloudflare: original announcement (2026-10-26). Sources checked November 3, 2026.