CVE-2026-105755: vLLM Cross-Request Integrity Break via Caller-Controlled Request ID

vllm Updated 06 Oct 2026

Summary

05/10/2026
Yes i
Moderate
4.2

CVE-2026-105755: Affected - Ecosystem / package: pip / vllm - Affected versions: vLLM ≤ 0.25.1 (confirmed on 0.25.1, commit 752a3a504485). Affects vllm versions < 0.30.0. Fixed in 0.30.0. How to fix ↓

Free scan

Check if this vulnerability affects your application

Upload a manifest or SBOM and we'll show whether you're affected, plus every other risky package in the same scan.

Check my app free

No signup required · Results in about 60 seconds

Upload your dependency manifest or SBOM

Drag & drop your file here or

Java Javascript PHP Python Go Rust Ruby Dart C/C++ Elixir CycloneDX SPDX

Upload a supported dependency manifest or SBOM. No installation or source-code access required.

See supported files and SBOM formats
SBOM sbom.json · bom.json · bom.xml · *.spdx.json · *.cdx.json · *.cdx.xml
Java pom.xml · gradle.lockfile · buildscript-gradle.lockfile
Javascript package-lock.json · yarn.lock · pnpm-lock.yaml
PHP composer.lock
Python poetry.lock · requirements.txt · Pipfile.lock
Go go.mod
Rust Cargo.lock
Ruby Gemfile.lock
Dart pubspec.lock
C/C++ conan.lock
Elixir mix.lock
Enter your work email to receive your report link.

Verify your email to view the full report and start 30 days of continuous monitoring. No credit card required.

Sponsored Lume, the Habilume companion Habilume Habit tracker with a companion that grows with you Complete habits, earn Lume Points, and watch your Lume grow as you stay consistent. App Store Google Play

Background

vLLM is a high-performance library designed for efficient large language model serving, enabling late-interaction scoring and reranking of queries. It is widely used in machine learning applications for its ability to handle concurrent requests efficiently. However, a vulnerability exists in versions up to 0.25.1, where the caching mechanism for query embeddings can be manipulated through a caller-controlled request ID, leading to potential data integrity issues.

Vulnerability Detail

The vulnerability arises from the way vLLM handles the caching of query embeddings during late-interaction scoring and reranking. Specifically, the system caches these embeddings using a key derived from the 'X-Request-Id' header, which can be controlled by the caller. This allows an attacker to reuse a victim's request ID, resulting in their query embedding overwriting the victim's cached entry. Consequently, the victim's documents are scored against the attacker's query, leading to a cross-request integrity break. Additionally, this can induce errors if the shared use counter is exhausted, forcing the victim's request into a cache-miss error. The vulnerability is confirmed in version 0.25.1 and potentially affects earlier versions.

How To Fix

Latest published version: 0.32.0 as of 22 Oct 2026

pip install --upgrade "vllm==0.32.0"

Upgrading to it may include breaking changes; the version below is the minimum fixed release for this vulnerability.

If you need the minimum change

0.30.0

This is the first release that contains the fix.

pip install --upgrade "vllm==0.30.0"

Tip — Not sure which version your app runs? Vulert can match this fix to it. Scan free →

Workaround

As a temporary workaround, you can disable flash late interaction if possible, or ensure that your application does not allow the reuse of 'X-Request-Id' values across different requests.

Conclusion

The vulnerability CVE-2026-105755 in vLLM poses significant risks to data integrity through cross-request manipulation of cached query embeddings. Users are advised to update to version 0.30.0 or later to mitigate these risks and ensure secure operation of their applications.

See every vulnerability recorded for vllm →

Share this vulnerability

FAQ

The vulnerability allows an attacker to corrupt another user's results by reusing their request ID, leading to incorrect scoring of documents.

You can use the Vulert playground at Vulert Scanner to check if your application is affected by this vulnerability without requiring any signup.

Versions of vLLM up to and including 0.25.1 are affected by this vulnerability.

Yes, a patch is available in version 0.30.0 and later. It is recommended to update to this version to mitigate the vulnerability.

The CVSS score of this vulnerability is 4.2, indicating a moderate severity level.

Related vulnerabilities

Recent vulnerabilities recorded for vllm.

Sponsored Lume, the Habilume companion Habilume Habit tracker with a companion that grows with you Complete habits, earn Lume Points, and watch your Lume grow as you stay consistent. App Store Google Play

How To Fix

Latest published version: 0.32.0 as of 22 Oct 2026

pip install --upgrade "vllm==0.32.0"

Upgrading to it may include breaking changes; the version below is the minimum fixed release for this vulnerability.

If you need the minimum change

0.30.0

This is the first release that contains the fix.

pip install --upgrade "vllm==0.30.0"

Tip — Not sure which version your app runs? Vulert can match this fix to it. Scan free →

Subscribe to Vulert on Telegram
Daily digest of new CVEs, critical and high severity, affected packages and fixes — plus a weekly roundup.
Join the channel →
Help improve this record
Report an issue or send feedback