EUVD-2026-82534

vLLM through 0.29.0 fails to properly clean up decode-side metadata for rejected inference requests in prefill/decode disaggregated deployments. Remote attackers can submit requests with maxtokens=0 to exhaust decode-worker memory without bound until the worker restarts...

CVE-2026-93436

vLLM through version 0.29.0 is vulnerable to a memory exhaustion issue in prefill/decode disaggregated deployments. The root cause is a failure to properly clean up decode-side metadata for rejected inference requests. A remote, unauthenticated attacker can submit requests with max_tokens=0 to ex...