Read the docs
TI-2026-38894 Aliases CVE-2026-53923GHSA-5jv2-g5wq-cmr4

vLLM GGUF Kernels: int64_t to int truncation of tensor dimensions causes GPU buffer overflow

Priority LOW 25CWE-200Published · Updated

Key facts

Attack vector
Network
Privileges required
None
User interaction
Passive
Attack complexity
Low
Attack requirements
None
Impact type
Information disclosure
Confidentiality · integrity · availability
Low · Low · None
Fix status
Fixed

Attack profile from CVSS 4.0 · CNA. Fix status from CNA, GitHub and Red Hat. Impact type from CWE-200.

Summary

vLLM is an inference and serving engine for large language models (LLMs). From 0.5.5 until 0.23.1rc0, integer truncation of tensor dimensions in vLLM's GGUF dequantize kernels (csrc/quantization/gguf/gguf_kernel.cu) causes partial tensor processing. The output tensor is allocated at full size via torch::empty (uninitialized memory), but the dequantize CUDA kernel processes only a truncated number of elements. The unfilled portion of the output tensor retains whatever was previously in GPU memory. In multi-tenant inference deployments, this residual GPU memory may contain tensor data from other users' inference requests, constituting information disclosure. This vulnerability is fixed in 0.23.1rc0.

Why it matters now

Exploitation None reported CISA SSVC · not in CISA KEV
EPSS 0.00484 41st percentileScored 2026-11-07
Fix Fixed vllm (vllm-project): 0.23.1rc0vllm (PyPI): 0.24.0Red Hat: 5 advisories

SSVC decision points

CISA ADP · SSVC 2.0.3

Exploitation
None
Automatable
No
Technical impact
Partial

The decision (Track, Attend, Act) also depends on your mission and well-being impact; see CISA's SSVC guide.

Affected products

VendorProductAffected versionsFixed inStatusSource
vllm-projectvllm0.5.5 to before 0.23.1rc00.23.1rc0Patch availableCNA
—vllm (PyPI)0.5.5 to before 0.24.00.24.0Patch availableGHSA-5jv2-g5wq-cmr4
  • vllm Vendorvllm-project Affected 0.5.5 to before 0.23.1rc0 Fixed in 0.23.1rc0 Patch available CNA
  • vllm (PyPI) Vendor— Affected 0.5.5 to before 0.24.0 Fixed in 0.24.0 Patch available GHSA-5jv2-g5wq-cmr4

CPE configurations NVD · 1

CPEVersionsRoleVendor
cpe:2.3:a:vllm:vllm:*:*:*:*:*:*:*:*0.5.5 to before 0.23.1VulnerablevLLM
  • cpe:2.3:a:vllm:vllm:*:*:*:*:*:*:*:* Versions 0.5.5 to before 0.23.1 Vulnerable ·vLLM

Remediation

  • Fix · CNA vllm (vllm-project) Fixed in 0.23.1rc0
  • Fix · GHSA-5jv2-g5wq-cmr4 vllm (PyPI) Fixed in 0.24.0 Advisory →

Red Hat 5 advisories

  • RHSA-2026:62336 Released 2026-09-01 · Red Hat rating Important
    Red Hat Enterprise Linux AI 3.3.6 enhancement update
    • Red Hat Enterprise Linux AI 3.3 registry.redhat.io/rhelai3/bootc-aws-cuda-rhel9, registry.redhat.io/rhelai3/bootc-azure-cuda-rhel9, registry.redhat.io/rhelai3/bootc-azure-rocm-rhel9 +3
  • RHSA-2026:62335 Released 2026-09-01 · Red Hat rating Important
    Red Hat Enterprise Linux AI 3.3.6 enhancement update
    • Red Hat Enterprise Linux AI 3.3 registry.redhat.io/rhelai3/disk-image-cuda-rhel9
  • RHSA-2026:60363 Released 2026-08-26 · Red Hat rating Important
    Red Hat AI Inference Server 3.3.6 (Spyre)
    • Red Hat AI Inference Server 3.3 registry.redhat.io/rhaiis/vllm-spyre-rhel9
Show 2 more Red Hat advisories
  • RHSA-2026:59139 Released 2026-08-24 · Red Hat rating Important
    Red Hat AI Inference Server 3.3.6 (ROCm)
    • Red Hat AI Inference Server 3.3 registry.redhat.io/rhaiis/vllm-rocm-rhel9
  • RHSA-2026:59138 Released 2026-08-24 · Red Hat rating Important
    Red Hat AI Inference Server 3.3.6 (CUDA)
    • Red Hat AI Inference Server 3.3 registry.redhat.io/rhaiis/vllm-cuda-rhel9

Vendor assessments

Red Hat

Red Hat rating: Low. Red Hat's page for this CVE →

ProductStatusPlan or fix
Red Hat AI Inference Server 3.3Fixed RHSA-2026:59138, RHSA-2026:59139, RHSA-2026:60363
Red Hat Enterprise Linux AI 3.3Fixed RHSA-2026:62335, RHSA-2026:62336
Not affected: 3 products
  • Red Hat AI Inference Server
  • Red Hat Enterprise Linux AI (RHEL AI) 3
  • Red Hat OpenShift AI (RHOAI)

Timeline

  1. 2026-10-02 Threatist record created from CVE-2026-53923
  2. 2026-10-03 Alias added: GHSA-5jv2-g5wq-cmr4

References

Patch 1

Advisory 3

Other 4

  • TI-2024-10229 Remote Code Execution in vllm-project/vllm CVSS 3.0CRITICAL 9.8
  • TI-2024-08404 Remote Code Execution in vllm-project/vllm CVSS 3.1CRITICAL 9.8
  • TI-2026-36017 vLLM: OpenAI auth bypass CVSS 3.1CRITICAL 9.1
  • TI-2026-63056 vLLM before 0.28.0 Denial of Service via negative token ID CVSS 4.0HIGH 8.7
  • TI-2026-04308 Hardcoded trust_remote_code=True in vllm-project/vllm Bypasses User Security Control CVSS 3.0HIGH 8.8
  • TI-2026-39137 vLLM: Remote DoS in vLLM via Invalid Recovered Token Reinjection CVSS 3.1HIGH 7.5
See all 124 →

Sources

This product uses the NVD API but is not endorsed or certified by the NVD. Sources and licences