PatchSiren

PatchSiren cyber security CVE debrief

CVE-2026-105755 vllm-project CVE debrief

A vulnerability in the vLLM package allows for a cross-request integrity break and induced errors on `/score` and `/rerank` due to the caching of query embeddings under a caller-controlled request ID. This issue affects vLLM versions ≤ 0.25.1 and is remotely reachable. The vulnerability arises from the worker caching per-request query embeddings under a key derived from the caller-controlled `X-Request-Id` header. A second concurrent request that reuses the victim's header value can replace the victim's cached query embedding before document scoring, leading to the victim's documents being scored against the attacker's query.

Vendor
vllm-project
Product
vllm
CVSS
MEDIUM 4.2
CISA KEV
Not listed in stored evidence
Original CVE published
2026-10-05
Original CVE updated
2026-10-08
Advisory published
2026-10-05
Advisory updated
2026-10-08

Who should care

Defenders responsible for deploying and managing vLLM should assess exposure and prioritize updates to address this vulnerability. They should review the usage of the `X-Request-Id` header and monitor for potential cache collisions and errors. Additionally, they should confirm whether affected product deployments exist in managed environments and assign an owner for follow-up.

Why it matters

Defenders should prioritize updating to version 0.30.0 or later to address this vulnerability, as it allows for a cross-request integrity break and induced errors on `/score` and `/rerank`. This issue affects vLLM versions ≤ 0.25.1 and is remotely reachable.

  • Potential for cross-request integrity breaks
  • Induced errors on `/score` and `/rerank`
  • Cache collisions and errors due to caller-controlled request IDs

Technical summary

The vLLM package caches query embeddings under a key derived from the caller-controlled `X-Request-Id` header. This allows for a cross-request integrity break and induced errors on `/score` and `/rerank`. The issue affects vLLM versions ≤ 0.25.1 and is remotely reachable. To address this vulnerability, defenders should prioritize updating to version 0.30.0 or later and review the usage of the `X-Request-Id` header to prevent cache collisions and errors. The vulnerability can be verified by checking for potential cache collisions and errors in the system.

Defensive priority

Defenders should prioritize updating to version 0.30.0 or later to address this vulnerability.

Recommended defensive actions

  • Update to version 0.30.0 or later
  • Review and adjust the usage of the `X-Request-Id` header
  • Monitor for potential cache collisions and errors
  • Confirm whether affected product deployments exist in managed environments and assign an owner for follow-up
  • Review compensating controls for exposed systems while remediation is scheduled and verified
  • Check relevant monitoring, detection, and logs for exposed assets that need extra review
  • Track exceptions, retest remediated assets, and close the item only after evidence is documented

Evidence notes

The vulnerability is caused by the caching of query embeddings under a key derived from the caller-controlled `X-Request-Id` header. A second concurrent request that reuses the victim's header value can replace the victim's cached query embedding before document scoring. This issue can be addressed by updating to version 0.30.0 or later and reviewing the usage of the `X-Request-Id` header.

Sources and references

Verified primary and authoritative sources

  • CVE-2026-105755 CVE Program record

    Publisher, destination, and source semantics verified

    URL: https://www.cve.org/CVERecord?id=CVE-2026-105755

    CVE Program - Official CVE Program record with source-provided CVE metadata.

  • CVE-2026-105755 NVD vulnerability detail

    Publisher, destination, and source semantics verified

    URL: https://nvd.nist.gov/vuln/detail/CVE-2026-105755

    NIST National Vulnerability Database - Official NIST NVD detail page and source-specific vulnerability assessment.

Supplemental references

Methodology and review provenance

AI-assisted synthesis based on stored public vulnerability evidence. System validation, approval state, and publication status do not by themselves establish human review of this revision. PatchSiren helps prioritize defensive review and does not prove exposure or remediation on any system.