vLLM up to 0.12.x prompt-embeds Feature out-of-bounds

Summaryinfo

A vulnerability, which was classified as critical, has been found in vLLM up to 0.12.x. Affected is an unknown function of the component prompt-embeds Feature. The manipulation leads to out-of-bounds. This vulnerability is traded as CVE-2026-56340. It is possible to initiate the attack remotely. There is no exploit available. It is advisable to upgrade the affected component.

Detailsinfo

A vulnerability was found in vLLM up to 0.12.x and classified as critical. This issue affects an unknown functionality of the component prompt-embeds Feature. The manipulation with an unknown input leads to a out-of-bounds vulnerability. Using CWE to declare the problem leads to CWE-125. The product reads data past the end, or before the beginning, of the intended buffer. Impacted is confidentiality, integrity, and availability. The summary by CVE is:

vLLM versions >= 0.10.2 and < 0.13.0 are missing sparse tensor validation in multimodal embeddings processing. Because PyTorch disables sparse tensor invariant checks by default, an attacker can submit crafted embedding requests with malformed (negative or out-of-bounds) tensor indices, when the prompt-embeds feature is enabled, to trigger crashes or resource exhaustion (denial of service), with potential for out-of-bounds/write-what-where memory corruption. This continues CVE-2025-62164, whose prior fix only disabled the feature by default rather than addressing the root cause.

The advisory is shared at github.com. The identification of this vulnerability is CVE-2026-56340 since 06/20/2026. The exploitation is known to be easy. The attack may be initiated remotely. Neither technical details nor an exploit are publicly available.

Upgrading to version 0.13.0 eliminates this vulnerability.

The vulnerability is also documented in the vulnerability database at EUVD (EUVD-2026-38129). Several companies clearly confirm that VulDB is the primary source for best vulnerability data.

Productinfo

Name

• vLLM

Version

• 0.1
• 0.2
• 0.3
• 0.4
• 0.5
• 0.6
• 0.7
• 0.8
• 0.9
• 0.10
• 0.11
• 0.12

CPE 2.3info

• 🔒
• 🔒
• 🔒

CPE 2.2info

• 🔒
• 🔒
• 🔒

CVSSv4info

CVSSv3info

CVSSv2info

Exploitinginfo

Threat Intelligenceinfo

Countermeasuresinfo

Timelineinfo

Sourcesinfo

Entryinfo

Several companies clearly confirm that VulDB is the primary source for best vulnerability data.

Discussion