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). 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 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). 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, 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 version 0.8.0 up to but excluding 0.9.0, the vLLM backend used with the /v1/chat/completions OpenAPI endpoint fails to validate unexpected or malformed input in the "pattern" and "type" fields when the tools functionality is invoked. These inputs are not validated before being compiled or parsed, causing a crash of the inference worker with a single request. The worker will remain down until it is restarted. Version 0.9.0 fixes the issue.

vLLM is a library for LLM inference and serving. From 0.12.0 to before 0.24.0, sending a pure prompt embeds payload in a /v1/completions request with a model using M-RoPE causes EngineCore to fail an assertion and fatally crash, shutting down the entire server application. Any remote user who is authorized to make a /v1/completions request can make such a request and induce a crash. This issue is fixed in version 0.24.0.

vLLM is an inference and serving engine for large language models (LLMs). Prior to 0.23.1rc0, vLLM's /v1/audio/transcriptions endpoint limits compressed upload size but not decoded PCM output. A 25MB OPUS file expands to ~14.9GB of float32 PCM at decode time. This vulnerability is fixed in 0.23.1rc0.

vLLM is an inference and serving engine for large language models (LLMs). From 0.18.0 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). 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 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 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 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 0.1.0 to before 0.10.1.1, a Denial of Service (DoS) vulnerability can be triggered by sending a single HTTP GET request with an extremely large header to an HTTP endpoint. This results in server memory exhaustion, potentially leading to a crash or unresponsiveness. The attack does not require authentication, making it exploitable by any remote user. This vulnerability is fixed in 0.10.1.1.

vLLM is a high-throughput and memory-efficient inference and serving engine for LLMs. Versions starting from 0.5.2 and prior to 0.8.5 are vulnerable to denial of service and data exposure via ZeroMQ on multi-node vLLM deployment. In a multi-node vLLM deployment, vLLM uses ZeroMQ for some multi-node communication purposes. The primary vLLM host opens an XPUB ZeroMQ socket and binds it to ALL interfaces. While the socket is always opened for a multi-node deployment, it is only used when doing tensor parallelism across multiple hosts. Any client with network access to this host can connect to this XPUB socket unless its port is blocked by a firewall. Once connected, these arbitrary clients will receive all of the same data broadcasted to all of the secondary vLLM hosts. This data is internal vLLM state information that is not useful to an attacker. By potentially connecting to this socket many times and not reading data published to them, an attacker can also cause a denial of service by slowing down or potentially blocking the publisher. This issue has been patched in version 0.8.5.

vLLM versions 0.8.0 and later are vulnerable to an Out-of-Memory (OOM) Denial of Service (DoS) attack due to unbounded frame count processing in the `VideoMediaIO.load_base64()` method. When processing `video/jpeg` data URLs, the method splits the base64 data string on commas to extract individual JPEG frames without enforcing a frame count limit. An attacker can exploit this by crafting a single API request containing thousands of comma-separated base64-encoded JPEG frames in a data URL, causing the server to decode all frames into memory and crash due to excessive memory consumption. This vulnerability is reachable via the OpenAI-compatible chat completions API and does not require authentication.

Mattermost fails to properly validate markdown, allowing an attacker to crash the server via a specially crafted markdown input.

An allocation of resources without limits or throttling vulnerability has been reported to affect Qsync Central. If a remote attacker gains a user account, they can then exploit the vulnerability to prevent other systems, applications, or processes from accessing the same type of resource. We have already fixed the vulnerability in the following version: Qsync Central 5.0.0.4 ( 2026/01/20 ) and later

In a Point-to-Multipoint (P2MP) Label Switched Path (LSP) scenario, an uncontrolled resource consumption vulnerability in the Routing Protocol Daemon (RPD) in Juniper Networks Junos OS allows a specific SNMP request to trigger an infinite loop causing a high CPU usage Denial of Service (DoS) condition. This issue affects both SNMP over IPv4 and IPv6. This issue affects: Juniper Networks Junos OS: 12.3X48 versions prior to 12.3X48-D90; 15.1 versions prior to 15.1R7-S6; 15.1X49 versions prior to 15.1X49-D200; 15.1X53 versions prior to 15.1X53-D238, 15.1X53-D592; 16.1 versions prior to 16.1R7-S5; 16.2 versions prior to 16.2R2-S11; 17.1 versions prior to 17.1R3-S1; 17.2 versions prior to 17.2R3-S2; 17.3 versions prior to 17.3R3-S7; 17.4 versions prior to 17.4R2-S4, 17.4R3; 18.1 versions prior to 18.1R3-S5; 18.2 versions prior to 18.2R3; 18.2X75 versions prior to 18.2X75-D50; 18.3 versions prior to 18.3R2; 18.4 versions prior to 18.4R2; 19.1 versions prior to 19.1R2.

Marinus Pfund, member of the AXIS OS Bug Bounty Program, has found the VAPIX API alwaysmulti.cgi was vulnerable for file globbing which could lead to resource exhaustion of the Axis device. Axis has released patched AXIS OS versions for the highlighted flaw. Please refer to the Axis security advisory for more information and solution.

A flaw was discovered in Wildfly's EJB Client as shipped with Red Hat JBoss EAP 7, where some specific EJB transaction objects may get accumulated over the time and can cause services to slow down and eventaully unavailable. An attacker can take advantage and cause denial of service attack and make services unavailable.

A vulnerability was found in wallabag 2.5.4. It has been declared as problematic. Affected by this vulnerability is an unknown functionality of the file /config of the component Profile Config. The manipulation of the argument Name leads to allocation of resources. The exploit has been disclosed to the public and may be used. The associated identifier of this vulnerability is VDB-233359. NOTE: The vendor was contacted early about this disclosure but did not respond in any way.

For GitLab before 13.0.12, 13.1.6, 13.2.3 a memory exhaustion flaw exists due to excessive logging of an invite email error message.

On BIG-IP Next CNF, BIG-IP Next SPK, and BIG-IP Next for Kubernetes systems, repeated undisclosed API calls can cause the Traffic Management Microkernel (TMM) to terminate.  Note: Software versions which have reached End of Technical Support (EoTS) are not evaluated.

Asterisk is an open source private branch exchange and telephony toolkit. Prior to versions 18.26.4 and 18.9-cert17, RTP UDP ports and internal resources can leak due to a lack of session termination. This could result in leaks and resource exhaustion. This issue has been patched in versions 18.26.4 and 18.9-cert17.

A vulnerability in the NETCONF over SSH feature of Cisco IOS XE Software could allow a low-privileged, authenticated, remote attacker to cause a denial of service condition (DoS) on an affected device. This vulnerability is due to insufficient resource management. An attacker could exploit this vulnerability by initiating a large number of NETCONF over SSH connections. A successful exploit could allow the attacker to exhaust resources, causing the device to reload and resulting in a DoS condition on an affected device.

An allocation of resources without limits or throttling vulnerability has been reported to affect File Station 5. If a remote attacker gains a user account, they can then exploit the vulnerability to prevent other systems, applications, or processes from accessing the same type of resource. We have already fixed the vulnerability in the following version: File Station 5 5.5.6.5018 and later

Zohocorp ManageEngine Exchange Reporter Plus through 5721 are vulnerable to ReDOS vulnerability in the search module.

File Browser provides a file managing interface within a specified directory and it can be used to upload, delete, preview, rename, and edit files. In version 2.38.0, a Denial of Service (DoS) vulnerability exists in the file processing logic when reading a file on endpoint `Filebrowser-Server-IP:PORT/files/{file-name}` . While the server correctly handles and stores uploaded files, it attempts to load the entire content into memory during read operations without size checks or resource limits. This allows an authenticated user to upload a large file and trigger uncontrolled memory consumption on read, potentially crashing the server and making it unresponsive. As of time of publication, no known patches are available.

An allocation of resources without limits or throttling vulnerability has been reported to affect File Station 5. If a remote attacker gains a user account, they can then exploit the vulnerability to prevent other systems, applications, or processes from accessing the same type of resource. We have already fixed the vulnerability in the following version: File Station 5 5.5.6.5018 and later

A denial of service vulnerability in B&R GateManager 4260 and 9250 versions <9.0.20262 and GateManager 8250 versions <9.2.620236042 allows authenticated users to limit availability of GateManager instances.

Windows Authentication Denial of Service Vulnerability

Allocation of Resources Without Limits or Throttling vulnerability in Apache Software Foundation Apache Struts.This issue affects Apache Struts: through 2.5.30, through 6.1.2. Upgrade to Struts 2.5.31 or 6.1.2.1 or greater.

An allocation of resources without limits or throttling vulnerability in the Schweitzer Engineering Laboratories SEL-451 could allow a remote authenticated attacker to make the system unavailable for an indefinite amount of time. See product Instruction Manual Appendix A dated 20230830 for more details.

Wyse Management Suite versions prior to 4.0 contain a denial-of-service vulnerability. An authenticated malicious user can flood the configured SMTP server with numerous requests in order to deny access to the system.

MeterSphere is an open source continuous testing platform. Version 2.9.1 and prior are vulnerable to denial of service. ​The `checkUserPassword` method is used to check whether the password provided by the user matches the password saved in the database, and the `CodingUtil.md5` method is used to encrypt the original password with MD5 to ensure that the password will not be saved in plain text when it is stored. If a user submits a very long password when logging in, the system will be forced to execute the long password MD5 encryption process, causing the server CPU and memory to be exhausted, thereby causing a denial of service attack on the server. This issue is fixed in version 2.10.0-lts with a maximum password length.

A resource exhaustion issue was addressed with improved input validation. This issue is fixed in macOS Big Sur 11.0.1. An attacker in a privileged network position may be able to perform denial of service.

An issue has been discovered in GitLab CE/EE affecting all versions before 17.10.7, 17.11 before 17.11.3, and 18.0 before 18.0.1. A lack of proper validation in GitLab could allow an authenticated user to cause a denial of service condition.

A denial of service exists in Gvisor Sandbox where a bug in reference counting code in mount point tracking could lead to a panic, making it possible for an attacker running as root and with permission to mount volumes to kill the sandbox. We recommend upgrading past commit 6a112c60a257dadac59962e0bc9e9b5aee70b5b6

IBM Sterling B2B Integrator 6.0.0.0 through 6.0.3.8 and 6.1.0.0 through 6.1.2.3 could allow an authenticated user to cause a denial of service due to uncontrolled resource consumption. IBM X-Force ID: 255827.

Due to an error in the software interface to the secure element chip on Bosch IP cameras of family CPP13 and CPP14, the chip can be permanently damaged when enabling the Stream security option (signing of the video stream) with option MD5, SHA-1 or SHA-256.

An uncontrolled resource consumption vulnerability has been reported to affect Qsync Central. If a remote attacker gains a user account, they can then exploit the vulnerability to launch a denial-of-service (DoS) attack. We have already fixed the vulnerability in the following version: Qsync Central 5.0.0.2 ( 2025/07/31 ) and later

Moxa IKS and EDS allow remote authenticated users to cause a denial of service via a specially crafted packet, which may cause the switch to crash.

IBM Security Access Manager Container (IBM Security Verify Access Appliance 10.0.0.0 through 10.0.6.1 and IBM Security Verify Access Docker 10.0.0.0 through 10.0.6.1) is vulnerable to a denial of service attacks on the DSC server. IBM X-Force ID: 254776.

Allocation of Resources Without Limits or Throttling vulnerability in Kron Technologies Kron PAM allows HTTP DoS. This issue affects Kron PAM: before 3.7.

IBM Db2 for Linux, UNIX and Windows (includes Db2 Connect Server) 10.5, 11.1, and 11.5 is vulnerable to denial of service with a specially crafted query.

SAP NetWeaver AS for ABAP (Business Server Pages) - versions 700, 701, 702, 731, 740, 750, 751, 752, 753, 754, 755, 756, 757, allows an attacker authenticated as a non-administrative user to craft a request with certain parameters in certain circumstances which can consume the server's resources sufficiently to make it unavailable over the network without any user interaction.

SAP NetWeaver AS for ABAP and ABAP Platform - versions 740, 750, 751, 752, 753, 754, 755, 756, 757, 791, allows an attacker authenticated as a non-administrative user to craft a request with certain parameters which can consume the server's resources sufficiently to make it unavailable over the network without any user interaction.