vLLM is an inference and serving engine for large language models (LLMs). From to before 0.20.0, the extract_hidden_states speculative decoding proposer in vLLM returns a tensor with an incorrect shape after the first decode step, causing a RuntimeError that crashes the EngineCore process. The crash is triggered when any request in the batch uses sampling penalty parameters (repetition_penalty, frequency_penalty, or presence_penalty). A single request with a penalty parameter (e.g., "repetition_penalty": 1.1) is sufficient to crash the server. This vulnerability is fixed in 0.20.0.

vLLM is an inference and serving engine for large language models (LLMs). From 0.6.1 to before 0.20.0, there is a a Token Injection vulnerability in vLLM’s multimodal processing. Unauthenticated, text-only prompts that spell special tokens are interpreted as control. Image and video placeholder sequences supplied without matching data cause vLLM to index into empty grids during input-position computation, raising an unhandled IndexError and terminating the worker or degrading availability. Multimodal paths that rely on image_grid_thw/video_grid_thw are affected. This vulnerability is fixed in 0.20.0.

vLLM is an inference and serving engine for large language models (LLMs). From 0.1.0 to before 0.19.0, a Denial of Service vulnerability exists in the vLLM OpenAI-compatible API server. Due to the lack of an upper bound validation on the n parameter in the ChatCompletionRequest and CompletionRequest Pydantic models, an unauthenticated attacker can send a single HTTP request with an astronomically large n value. This completely blocks the Python asyncio event loop and causes immediate Out-Of-Memory crashes by allocating millions of request object copies in the heap before the request even reaches the scheduling queue. This vulnerability is fixed in 0.19.0.

vLLM is an inference and serving engine for large language models (LLMs). From 0.7.0 to before 0.19.0, the VideoMediaIO.load_base64() method at vllm/multimodal/media/video.py splits video/jpeg data URLs by comma to extract individual JPEG frames, but does not enforce a frame count limit. The num_frames parameter (default: 32), which is enforced by the load_bytes() code path, is completely bypassed in the video/jpeg base64 path. An attacker can send a single API request containing thousands of comma-separated base64-encoded JPEG frames, causing the server to decode all frames into memory and crash with OOM. This vulnerability is fixed in 0.19.0.

vLLM is a high-throughput and memory-efficient inference and serving engine for LLMs. The outlines library is one of the backends used by vLLM to support structured output (a.k.a. guided decoding). Outlines provides an optional cache for its compiled grammars on the local filesystem. This cache has been on by default in vLLM. Outlines is also available by default through the OpenAI compatible API server. The affected code in vLLM is vllm/model_executor/guided_decoding/outlines_logits_processors.py, which unconditionally uses the cache from outlines. A malicious user can send a stream of very short decoding requests with unique schemas, resulting in an addition to the cache for each request. This can result in a Denial of Service if the filesystem runs out of space. Note that even if vLLM was configured to use a different backend by default, it is still possible to choose outlines on a per-request basis using the guided_decoding_backend key of the extra_body field of the request. This issue applies only to the V0 engine and is fixed in 0.8.0.

vLLM is an inference and serving engine for large language models (LLMs). In versions from 0.6.4 to before 0.12.0, users can crash the vLLM engine serving multimodal models that use the Idefics3 vision model implementation by sending a specially crafted 1x1 pixel image. This causes a tensor dimension mismatch that results in an unhandled runtime error, leading to complete server termination. This issue has been patched in version 0.12.0.

vLLM is an inference and serving engine for large language models (LLMs). From version 0.5.5 to before 0.11.1, the /v1/chat/completions and /tokenize endpoints allow a chat_template_kwargs request parameter that is used in the code before it is properly validated against the chat template. With the right chat_template_kwargs parameters, it is possible to block processing of the API server for long periods of time, delaying all other requests. This issue has been patched in version 0.11.1.

vLLM is an inference and serving engine for large language models (LLMs). From version 0.5.5 to before 0.11.1, users can crash the vLLM engine serving multimodal models by passing multimodal embedding inputs with correct ndim but incorrect shape (e.g. hidden dimension is wrong), regardless of whether the model is intended to support such inputs (as defined in the Supported Models page). This issue has been patched in version 0.11.1.

vLLM, an inference and serving engine for large language models (LLMs), has a Regular Expression Denial of Service (ReDoS) vulnerability in the file `vllm/entrypoints/openai/tool_parsers/pythonic_tool_parser.py` of versions 0.6.4 up to but excluding 0.9.0. The root cause is the use of a highly complex and nested regular expression for tool call detection, which can be exploited by an attacker to cause severe performance degradation or make the service unavailable. The pattern contains multiple nested quantifiers, optional groups, and inner repetitions which make it vulnerable to catastrophic backtracking. Version 0.9.0 contains a patch for the issue.

vLLM is an inference and serving engine for large language models (LLMs). In versions 0.8.0 up to but excluding 0.9.0, hitting the /v1/completions API with a invalid json_schema as a Guided Param kills the vllm server. This vulnerability is similar GHSA-9hcf-v7m4-6m2j/CVE-2025-48943, but for regex instead of a JSON schema. Version 0.9.0 fixes the issue.

vLLM is an inference and serving engine for large language models (LLMs). Version 0.8.0 up to but excluding 0.9.0 have a Denial of Service (ReDoS) that causes the vLLM server to crash if an invalid regex was provided while using structured output. This vulnerability is similar to GHSA-6qc9-v4r8-22xg/CVE-2025-48942, but for regex instead of a JSON schema. Version 0.9.0 fixes the issue.

vLLM is a high-throughput and memory-efficient inference and serving engine for LLMs. Versions starting from 0.8.0 and prior to 0.8.5 are affected by a critical performance vulnerability in the input preprocessing logic of the multimodal tokenizer. The code dynamically replaces placeholder tokens (e.g., <|audio_|>, <|image_|>) with repeated tokens based on precomputed lengths. Due to ​​inefficient list concatenation operations​​, the algorithm exhibits ​​quadratic time complexity (O(n²))​​, allowing malicious actors to trigger resource exhaustion via specially crafted inputs. This issue has been patched in version 0.8.5.

vLLM is an inference and serving engine for large language models (LLMs). From version 0.5.5 to before version 0.18.0, Librosa defaults to using numpy.mean for mono downmixing (to_mono), while the international standard ITU-R BS.775-4 specifies a weighted downmixing algorithm. This discrepancy results in inconsistency between audio heard by humans (e.g., through headphones/regular speakers) and audio processed by AI models (Which infra via Librosa, such as vllm, transformer). This issue has been patched in version 0.18.0.

vLLM is an inference and serving engine for large language models (LLMs). From versions 0.10.2 to before 0.11.1, a memory corruption vulnerability could lead to a crash (denial-of-service) and potentially remote code execution (RCE), exists in the Completions API endpoint. When processing user-supplied prompt embeddings, the endpoint loads serialized tensors using torch.load() without sufficient validation. Due to a change introduced in PyTorch 2.8.0, sparse tensor integrity checks are disabled by default. As a result, maliciously crafted tensors can bypass internal bounds checks and trigger an out-of-bounds memory write during the call to to_dense(). This memory corruption can crash vLLM and potentially lead to code execution on the server hosting vLLM. This issue has been patched in version 0.11.1.

The Reviews Plus WordPress plugin before 1.2.14 does not validate the submitted rating, allowing submission of long integer, causing a Denial of Service in the review section when an authenticated user submit such rating and the reviews are set to be displayed on the post/page

PwnDoc is a penetration test report generator. In versions up to and including 0.5.3, an authenticated user is able to crash the backend by raising a `UnhandledPromiseRejection` on audits which exits the backend. The user doesn't need to know the audit id, since a bad audit id will also raise the rejection. With the backend being unresponsive, the whole application becomes unusable for all users of the application. As of time of publication, no known patches are available.

A flaw was found in Red Hat 3scale API Management Platform 2. The 3scale backend does not perform preventive handling on user-requested date ranges in certain queries allowing a malicious authenticated user to submit a request with a sufficiently large date range to eventually yield an internal server error resulting in denial of service. The highest threat from this vulnerability is to system availability.

A user authorized to performing a specific type of find query may trigger a denial of service. This issue affects MongoDB Server v4.4 versions prior to 4.4.4.

Multiple vulnerabilities in Cisco Jabber for Windows, Cisco Jabber for Mac, and Cisco Jabber for mobile platforms could allow an attacker to access sensitive information or cause a denial of service (DoS) condition. For more information about these vulnerabilities, see the Details section of this advisory.

A vulnerability in the API of Cisco Meeting Server could allow an authenticated, remote attacker to cause a denial of service (DoS) condition on an affected device. This vulnerability exists because requests that are sent to the API are not properly validated. An attacker could exploit this vulnerability by sending a malicious request to the API. A successful exploit could allow the attacker to cause all participants on a call to be disconnected, resulting in a DoS condition.

A vulnerability has been identified in SIMATIC CN 4100 (All versions < V4.0). The affected application allows to control the device by storing arbitrary files in the SFTP folder of the device. This could allow an attacker to cause a denial of service condition.

A vulnerability has been identified in SINEC Security Monitor (All versions < V4.10.0). The affected application lacks input validation of date parameter in report generation functionality. This could allow an authenticated, lowly privileged attacker to cause denial of service condition of the report functionality.

CWE-20: Improper Input Validation vulnerability exists that could cause Denial of Service when an authenticated malicious user sends HTTPS request containing invalid data type to the webserver.

Insufficient Input Validation in Web Applications operating on Business-DNA Solutions GmbH’s TopEase® Platform Version <= 7.1.27 on all object attributes allows an authenticated remote attacker with Object Modification privileges to insert arbitrarily long strings, eventually leading to exhaustion of the underlying resource.

Missing input validation in the MP_REACH_NLRI component of FRRouting (FRR) stable/10.0 to stable/10.6 allows authenticated attackers to cause a Denial of Service (DoS) via supplying a crafted UPDATE message.

In Apache Subversion versions up to and including 1.9.10, 1.10.4, 1.12.0, Subversion's svnserve server process may exit when a well-formed read-only request produces a particular answer. This can lead to disruption for users of the server.

Improper input validation vulnerability in SettingsProvider prior to Android S(12) allows privileged attackers to trigger a permanent denial of service attack on a victim's devices.

Nginx UI is a web user interface for the Nginx web server. Prior to version 2.3.4, an input validation vulnerability in the logrotate configuration allows an authenticated user to cause a complete Denial of Service (DoS). By submitting a negative integer for the rotation interval, the backend enters an infinite loop or an invalid state, rendering the web interface unresponsive. This issue has been patched in version 2.3.4.

Vulnerability in the MySQL Server component of Oracle MySQL (subcomponent: Server: DDL). Supported versions that are affected are 5.5.53 and earlier, 5.6.34 and earlier and 5.7.16 and earlier. Easily exploitable vulnerability allows low privileged attacker with network access via multiple protocols to compromise MySQL Server. Successful attacks of this vulnerability can result in unauthorized ability to cause a hang or frequently repeatable crash (complete DOS) of MySQL Server. CVSS v3.0 Base Score 6.5 (Availability impacts).

Kirby CMS through 5.1.4 allows an authenticated user with 'Editor' permissions to cause a persistent Denial of Service (DoS) via a malformed image upload. The application fails to properly validate the return value of the PHP getimagesize() function. When the system attempts to process this file for metadata or thumbnail generation, it triggers a fatal TypeError.

CWE-20: Improper Input Validation vulnerability exists that could cause Denial of Service when an authenticated malicious user sends special malformed HTTPS request containing improper formatted body data to the controller.

Improper Input Validation vulnerability in The Wikimedia Foundation Mediawiki - GrowthExperiments allows HTTP DoS.This issue affects Mediawiki - GrowthExperiments: from 1.39 through 1.43.

Several denial of service vulnerabilities exist in Eternal Terminal prior to version 6.2.0, including a DoS triggered remotely by an invalid sequence number and a local bug triggered by invalid input sent directly to the IPC socket.

Improper input validation in Active Directory Certificate Services (AD CS) allows an authorized attacker to deny service over a network.

A Denial of Service vulnerability was found in Apache Qpid Dispatch Router versions 0.7.0 and 0.8.0. To exploit this vulnerability, a remote user must be able to establish an AMQP connection to the Qpid Dispatch Router and send a specifically crafted AMQP frame which will cause it to segfault and shut down.

Discourse is an option source discussion platform. Prior to version 2.8.14 on the `stable` branch and version 2.9.0.beta16 on the `beta` and `tests-passed` branches, users can create posts with raw body longer than the `max_length` site setting by including html comments that are not counted toward the character limit. This issue is patched in versions 2.8.14 and 2.9.0.beta16. There are no known workarounds.

It is possible to cause a DoS condition by causing the server to crash in alien-arena 7.33 by supplying various invalid parameters to the download command.

A vulnerability in the management REST API of Cisco Industrial Network Director (IND) could allow an authenticated, remote attacker to cause the CPU utilization to increase to 100 percent, resulting in a denial of service (DoS) condition on an affected device. The vulnerability is due to insufficient validation of requests sent to the REST API. An attacker could exploit this vulnerability by sending a crafted request to the REST API. A successful exploit could allow the attacker to cause a permanent DoS condition that is due to high CPU utilization. Manual intervention may be required to recover the Cisco IND.

Improper input validation in the Netconf service in Infinera MTC-9 allows remote authenticated users to crash the service and reboot the appliance, thus causing a DoS condition, via crafted XML payloads.This issue affects MTC-9: from R22.1.1.0275 before R23.0.

When BIG-IP Next Central Manager is running, undisclosed requests to the BIG-IP Next Central Manager API can cause the BIG-IP Next Central Manager Node's Kubernetes service to terminate. Note: Software versions which have reached End of Technical Support (EoTS) are not evaluated.

Synapse is a Matrix reference homeserver written in python (pypi package matrix-synapse). Matrix is an ecosystem for open federated Instant Messaging and VoIP. In Synapse before version 1.28.0 Synapse is missing input validation of some parameters on the endpoints used to confirm third-party identifiers could cause excessive use of disk space and memory leading to resource exhaustion. Note that the groups feature is not part of the Matrix specification and the chosen maximum lengths are arbitrary. Not all clients might abide by them. Refer to referenced GitHub security advisory for additional details including workarounds.

Synapse is a Matrix reference homeserver written in python (pypi package matrix-synapse). Matrix is an ecosystem for open federated Instant Messaging and VoIP. In Synapse before version 1.28.0 Synapse is missing input validation of some parameters on the endpoints used to confirm third-party identifiers could cause excessive use of disk space and memory leading to resource exhaustion. Note that the groups feature is not part of the Matrix specification and the chosen maximum lengths are arbitrary. Not all clients might abide by them. Refer to referenced GitHub security advisory for additional details including workarounds.

In Splunk Enterprise versions below 10.0.2, 9.4.6, 9.3.8, and 9.2.10, and versions below 3.9.10, 3.8.58 and 3.7.28 of the Splunk Secure Gateway app on Splunk Cloud Platform, a low-privileged user that does not hold the "admin" or "power" Splunk roles could craft a malicious payload through the `label` column field after adding a new device in the Splunk Secure Gateway app. This could potentially lead to a client-side denial of service (DoS).

An attacker with basic CRUD permissions on a replicated collection can run the applyOps command with specially malformed oplog entries, resulting in a potential denial of service on secondaries. This issue affects MongoDB Server v4.0 versions prior to 4.0.27; MongoDB Server v4.2 versions prior to 4.2.16; MongoDB Server v4.4 versions prior to 4.4.9.

A flaw (CVE-2022-38900) was discovered in one of Kibana’s third party dependencies, that could allow an authenticated user to perform a request that crashes the Kibana server process.

A Denial of Service vulnerability exists in the ITMS workflow process manager login window in Symantec IT Management Suite 8.0.

Cube is a semantic layer for building data applications. Prior to version 0.34.34, it is possible to make the entire Cube API unavailable by submitting a specially crafted request to a Cube API endpoint. The issue has been patched in `v0.34.34` and it's recommended that all users exposing Cube APIs to the public internet upgrade to the latest version to prevent service disruption. There are currently no workaround for older versions, and the recommendation is to upgrade.

On versions 16.1.x before 16.1.2 and 15.1.x before 15.1.4.1, when BIG-IP APM portal access is configured on a virtual server, undisclosed requests can cause the Traffic Management Microkernel (TMM) to terminate. Note: Software versions which have reached End of Technical Support (EoTS) are not evaluated.

qpid-cpp 1.0 crashes when a large message is sent and the Digest-MD5 mechanism with a security layer is in use .

An inaccurate frame deduplication process in ChirpStack Network Server 3.9.0 allows a malicious gateway to perform uplink Denial of Service via malformed frequency attributes in CollectAndCallOnceCollect in internal/uplink/collect.go. NOTE: the vendor's position is that there are no "guarantees that allowing untrusted LoRa gateways to the network should still result in a secure network.