CVE-2026-56340Disclosure(vllm / vllm)

LOWCVSS 7.5 · HIGH

Signal is active with 1 mentions in latest observed window

Immediate actions

Track advisory updates for patch or workaround availability

Recommended action window: Monitor and triage in normal cycle

NVD description

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.

vllm CVEs Back to dashboard

0.0/ 10 priority

Sources & remediation

Vendor / third-party advisories

Other references

Weakness type (CWE)

CWE-20CWE-787

Priority

LOW

Exploitation

NONE

PoC

NONE

Patch

NONE

Momentum

STABLE

Are you affected?

If you run products in this scope, you should treat this CVE as relevant to your environment.

vllm

Threat summary

4 mentions across 2 observed days
Momentum state: stable

What's happening

Technical details provided in 4 signals
Disclosure: 4 classified signals
Peaked 1d ago at 3 mentions (2026-06-20); latest day: 1
4 total mentions across 2 days

Affected systems

Vendors

vllm

Products

vllm

Deep dive

Activity timeline4 mentions / 2d

Signal classification1 categories

Disclosure

4100.0%

Referenced assets3 URLs

By indicator

Classification over time

Date	Total	Labels

2026-06-20	3	Disclosure3
2026-06-21	1	Disclosure1

Full discourse4 posts

Upwind Security MDR@UpwindMDR91d
Disclosure
🚨 HIGH - vLLM prompt-embeds sparse tensor validation flaw enables DoS/memory corruption (CVE-2026-56340) vLLM fails to properly validate sparse tensor indices during multimodal embeddings processing when the prompt-embeds feature is enabled. The root cause is missing bounds and consistency checks on attacker-controlled sparse tensor metadata, allowing malformed indices to reach lower-level tensor operations. An attacker can exploit this by sending crafted embedding requests that trigger crashes or sustained resource exhaustion, resulting in denial of service. In worst cases, out-of-bounds behavior may lead to write-what-where style memory corruption, and the risk is amplified because a prior related CVE only disabled the feature by default instead of fixing validation. 👉 Affected: vllm (prompt-embeds enabled; versions not specified) | No fix yet - treat as suspicious
Post summary
A high‑severity vulnerability (CVE‑2026‑56340) in vLLM’s prompt‑embeds feature allows attackers to trigger DoS or out‑of‑bounds memory corruption by supplying malformed sparse tensor indices. No patch or workaround is currently available.
0000071
223 followersView on X
Vulmon Vulnerability Feed@VulmonFeeds91d
Disclosure
CVE-2026-56340 Sparse Tensor Validation Missing in vLLM Multimodal Embeddings Processing https://vulmon.com/vulnerabilitydetails?qid=CVE-2026-56340
Post summary
CVE-2026-56340, a missing sparse tensor validation flaw in vLLM's multimodal embeddings processing, is disclosed without any PoC, exploit, or patch information.
0000054
4.1K followersView on X
Infoflowcloud@infoflowcloud91d
Disclosure
🚨*CVE* CVE-2026-56340 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… https://www.cve.org/CVERecord?id=CVE-2026-56340 ----- Traducción: CVE-2026-56340 vLL… http://infoflow.cloud`
Post summary
The post announces CVE-2026-56340, detailing that vLLM versions 0.10.2 through <0.13.0 lack sparse tensor validation in multimodal embeddings, and provides a link to the official CVE record.
0000037
88 followersView on X
CVE@CVEnew91d
Disclosure
CVE-2026-56340 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… https://www.cve.org/CVERecord?id=CVE-2026-56340
Post summary
The post reports that vLLM versions 0.10.2–0.12.x lack proper sparse tensor validation in multimodal embeddings, referencing the CVE record for more detail.
00000338
57.7K followersView on X

CPE platform detail1 entries

Part:

Vendor:

Product:

Version:

Target SW:

Target HW:

1 of 1 entries

Part	Vendor	Product	Version	Target SW	Target HW
App	vllm	vllm	-	-	-

Explore more

Back to dashboard vllm CVEs vllm vulnerabilities

What researchers are saying

CVE-2026-56340

Upwind Security MDR[verified]@UpwindMDRDisclosure

A high‑severity vulnerability (CVE‑2026‑56340) in vLLM’s prompt‑embeds feature allows attackers to trigger DoS or out‑of‑bounds memory corruption by supplying malformed sparse tensor indices. No patch or workaround is currently available.

Vulmon Vulnerability Feed@VulmonFeedsDisclosure

CVE-2026-56340, a missing sparse tensor validation flaw in vLLM's multimodal embeddings processing, is disclosed without any PoC, exploit, or patch information.

Infoflowcloud@infoflowcloudDisclosure

The post announces CVE-2026-56340, detailing that vLLM versions 0.10.2 through <0.13.0 lack sparse tensor validation in multimodal embeddings, and provides a link to the official CVE record.

CVE@CVEnewDisclosure

The post reports that vLLM versions 0.10.2–0.12.x lack proper sparse tensor validation in multimodal embeddings, referencing the CVE record for more detail.