vulnti.work

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 max_tokens=0…

High CVSS 7.5
CVECVE-2026-93436
First seen2026-09-22 04:15 UTC
Disclosed2026-09-17 23:18 UTC
Last updated2026-09-22 04:15 UTC
Channel statusauto

Summary

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 max_tokens=0 to exhaust decode-worker memory without bound until the worker restarts.

In-depth triage

No in-depth report has been generated yet (DR-003 v2 AI pipeline is under construction).

Sources

  • NVD DATABASE

Original Links

Timeline

  1. nvd_ingest NVD