Vulnerability Description
vLLM is an inference and serving engine for large language models. Prior to 0.32.0, an authenticated caller can place forged shared-memory address and monotonic_id values in public multimodal features.kwargs_data. In scale-out deployments with shared-memory multimodal caching enabled, workers trust those values and pass attacker-selected bytes to pickle.loads(), allowing arbitrary Python execution even when insecure serialization is disabled. This issue is fixed in version 0.32.0.
CVSS Score
HIGH
Affected Products
| Vendor | Product | Versions |
|---|---|---|
| Vllm | Vllm | < 0.32.0 |
Related Weaknesses (CWE)
References
- https://github.com/vllm-project/vllm/commit/3d5625cc3c2474994107047d7a6ecb8e4008Patch
- https://github.com/vllm-project/vllm/pull/59357Patch
- https://github.com/vllm-project/vllm/releases/tag/v0.32.0Release Notes
- https://github.com/vllm-project/vllm/security/advisories/GHSA-6757-hggm-hp2wPatchVendor Advisory
FAQ
What is CVE-2026-115912?
CVE-2026-115912 is a vulnerability with a CVSS score of 7.5 (HIGH). vLLM is an inference and serving engine for large language models. Prior to 0.32.0, an authenticated caller can place forged shared-memory address and monotonic_id values in public multimodal features...
How severe is CVE-2026-115912?
CVE-2026-115912 has been rated HIGH with a CVSS base score of 7.5/10. Review the CVSS metrics above for detailed severity breakdown.
Is there a patch for CVE-2026-115912?
Check the references section above for vendor advisories and patch information. Affected products include: Vllm Vllm.