CVE-2026-114049
LOW
NVD
CVSS Score
3.7
Severity
LOW
Published
Oct 21, 2026
Vendor
unknown
Description
vLLM before 0.31.0 contains a cross-request mask leak in InputBatch.condense() that allows remote attackers to constrain other clients' outputs by leaving stale allowed_token_ids rows. Attackers can submit concurrent constrained generate requests, including via unauthenticated /invocations, so recycled batch slots apply their token allowlist to victims' unrestricted requests.