Medium severity5.3NVD Advisory· Published Aug 13, 2026· Updated Sep 28, 2026
CVE-2026-73558
CVE-2026-73558
Description
vLLM is an inference and serving engine for large language models. Prior to 0.27.0, an integer overflow in blockIdx.x * 2 * d in activation_kernels.cu can cause act_and_mul_kernel to consume another batched user's input, allowing a request processed in the same inference batch to receive a partial or complete copy of another user's inference result. This issue is fixed in version 0.27.0.
AI Insight
LLM-synthesized narrative grounded in this CVE's description and references.
Affected packages
Versions sourced from the GitHub Security Advisory.
| Package | Affected versions | Patched versions |
|---|---|---|
vllmPyPI | < 0.27.0 | 0.27.0 |
Affected products
2Patches
Vulnerability mechanics
References
7- github.com/vllm-project/vllm/commit/451227cb3ff07989698fed982c2d3e4300257924nvdPatchWEB
- github.com/vllm-project/vllm/issues/42860nvdIssue TrackingPatchWEB
- github.com/vllm-project/vllm/pull/49660nvdIssue TrackingPatchWEB
- github.com/vllm-project/vllm/security/advisories/GHSA-7m6h-x95x-82q5nvdExploitVendor AdvisoryWEB
- github.com/advisories/GHSA-7m6h-x95x-82q5ghsaADVISORY
- nvd.nist.gov/vuln/detail/CVE-2026-73558ghsaADVISORY
- github.com/vllm-project/vllm/releases/tag/v0.27.0nvdRelease NotesWEB
News mentions
0No linked articles in our index yet.