Vulnerability in Vllm
CVE-2026-94627
vLLM Mooncake connector through 0.29.0 fails to properly manage GPU KV cache block ownership when concurrent child requests share a single transfer ID in prefill/decode disaggregated deployments. Attackers can trigger GPU memory exhaustion by submitting completion requests with multiple prompts, causing orphaned KV cache blocks to accumulate until process restart and eventually preventing legitimate requests from executing.
Published · last modified .
CVSS v3 metric
CVSS v3 base score 7.5 (High). Vector: CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H.
EPSS exploit prediction
EPSS: 0.006 (50.1th percentile), scored .
Very low probability of exploitation in the next 30 days; routine patching cadence is appropriate. 50th percentile — 50.1% of CVEs in the catalogue have a lower EPSS than this one. How to read EPSS.
EPSS over last 30 days · oldest: 0.006 · newest: 0.006 · change: 0.000
Affected products
Weakness classification (CWE)
References
- github.com/vllm-project/vllm (Product, product)
- github.com/vllm-project/vllm/blob/v0.29.0/vllm/distributed/kv_transfer/kv_conne… (Product, technical-description)
- github.com/vllm-project/vllm/pull/49796 (issue-tracking, Patch, patch, Issue Tracking)
- www.vulncheck.com/advisories/vllm-through-0.29.0-gpu-kv-cache-leak-via-mooncake… (Third Party Advisory, third-party-advisory)
Frequently asked questions
- What is CVE-2026-94627?
- CVE-2026-94627 is a high-severity vulnerability in Vllm, classified under Missing Release of Memory after Effective Lifetime. CVSS score: 7.5/10. Published 2026-09-21.
- How severe is CVE-2026-94627?
- High severity. CVSS v3 base score is 7.5 out of 10.