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.
CVE-2026-93436
Score 7.5 from GitHub Security Advisory (severity: HIGH) published 2026-09-18. a secondary CVSS source baseline 7.5; sources differ by 0.0.
- High severity, but no confirmed exploitation yet
No vendor fix yet — apply a workaround or compensating control (WAF / firewall / segmentation) and watch for a patch.
- CVSS v3
- 7.5
- EG Score
- 7.5(high)
- EG Risk
- 38(Track)EG Risk 38/100SSVC: Track
EG Risk is EchelonGraph's 0–100 priority score: it fuses intrinsic severity with real-world exploitation and automatability so you can rank equal-severity CVEs and fix the most dangerous first. Higher = act sooner. Distinct from the 0–10 EG Score (severity).
How it’s computedSeverity75% × 45%Exploitation1% × 40%Automatability30% × 15%Action: Routine — remediate on your standard cadence. - EPSS PROB
- 1%
- EPSS %ILE
- 44%
- KEV
- Not listed
Published
September 17, 2026
Last Modified
September 17, 2026
Advisory Details (4)
Auto-updated Sep 17, 2026vLLM through 0.29.0 Memory Exhaustion via Rejected Requests | Advisories | VulnCheck
https://www.vulncheck.com/advisories/vllm-through-0.29.0-memory-exhaustion-via-rejected-requestsGitHub - vllm-project/vllm: A high-throughput and memory-efficient inference and serving engine for LLMs · GitHub
https://github.com/vllm-project/vllmvllm/vllm/distributed/kv_transfer/kv_connector/v1/nixl/push_worker.py at v0.29.0 · vllm-project/vllm · GitHub
https://github.com/vllm-project/vllm/blob/v0.29.0/vllm/distributed/kv_transfer/kv_connector/v1/nixl/push_worker.py#L162-L181[Security] Drop leftover NIXL empty-recv metadata
Upstream fix in progress (PR #55677 · vllm-project/vllm, state=open)
https://github.com/vllm-project/vllm/pull/55677Vendor Advisories for CVE-2026-93436(1)
These vendors published their own advisory mentioning this CVE — often with vendor-specific remediation steps + affected product lists not in NVD.
Weakness Classification(1)
MITRE Common Weakness Enumeration — the root-cause categories this CVE belongs to.
Data Freshness Timeline
(refreshed 12× in last 7d / 12× in last 30d)
Each row is a source pipeline that fetched or updated this CVE on that date, with what changed. For example, "NVD update" means NVD published or revised its analysis for this CVE; "MITRE cvelistV5" means we ingested or refreshed it from the CNA feed. Most recent first.
- 2026-09-20 05:53 UTCGHSA enrichment
- 2026-09-19 19:00 UTCEG score recompute
- 2026-09-19 19:00 UTCGHSA enrichment
- 2026-09-19 08:06 UTCGHSA enrichment
- 2026-09-18 21:13 UTCEG score recompute
- 2026-09-18 21:13 UTCGHSA enrichment
- 2026-09-18 19:28 UTCEPSS rescore
- 2026-09-18 10:19 UTCEG score recompute
- 2026-09-18 10:19 UTCGHSA enrichment
- 2026-09-17 23:25 UTCEG score recompute
- 2026-09-17 22:25 UTCEG score recompute
- 2026-09-17 22:24 UTCMITRE cvelistV5first tracked
Related CVEs(same CWE)
Frequently asked(5)
What is CVE-2026-93436?
When was CVE-2026-93436 disclosed?
Is CVE-2026-93436 actively exploited?
What is the CVSS score of CVE-2026-93436?
How do I remediate CVE-2026-93436?
Dependency Blast Radius
Explore the affected products and dependency analysis for CVE-2026-93436
Is Your Infrastructure Affected by CVE-2026-93436?
EchelonGraph automatically scans your cloud infrastructure and maps CVE exposure using blast radius analysis.