Information disclosure in Vllm
CVE-2026-105752
- CVSS v3
- 3.1 Low
- Public exploits
- None known
- EPSS
- 0.002 · 12.3th percentile
- Published
- 2026-10-05
What to do
- Fixed in
0.30.0- Affected
- 2 affected products
vLLM is an inference and serving engine for large language models. Prior to 0.30.0, Harmony tool continuations submitted through "POST /v1/responses" requests rebuild the next-turn engine input without preserving the cache_salt value, placing the continuation prefix in the global unsalted cache namespace even when the caller enabled salting. On deployments with prefix caching enabled, which is the default, an authenticated tenant who can reconstruct a victim's low-entropy post-tool history can submit the same continuation and use the cached_tokens_per_turn count to determine whether the prefix was previously processed, defeating the intended tenant isolation of salted prefix caching. This issue is fixed in version 0.30.0.
Vulnerability class: Information Disclosure
Published · last modified .
CVSS v3 metric
CVSS v3 base score 3.1 (Low). Vector: CVSS:3.1/AV:N/AC:H/PR:L/UI:N/S:U/C:N/I:L/A:N.
- Attack Vector
- Network
AV:N - Attack Complexity
- High
AC:H - Privileges Required
- Low
PR:L - User Interaction
- None
UI:N - Scope
- Unchanged
S:U - Confidentiality
- None
C:N - Integrity
- Low
I:L - Availability
- None
A:N
EPSS exploit prediction
EPSS 0.002, 12.3th percentile (last changed ); unchanged over 30 days. About EPSS.
Affected products
- Vllm
- Vllm-Project Vllm — versions < 0.30.0
Weakness classification (CWE)
References
- https://github.com/vllm-project/vllm/security/advisories/GHSA-935w-9g4m-p28p (Mitigation, Vendor Advisory)
- https://github.com/vllm-project/vllm/pull/50195 (Patch, Issue Tracking)
- https://github.com/vllm-project/vllm/pull/51818 (Patch, Issue Tracking)
- https://github.com/vllm-project/vllm/commit/6a2a2bb02b563b83f946012959fd3927984d072a (Patch)
- https://github.com/vllm-project/vllm/releases/tag/v0.30.0 (Release Notes)