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, 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 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. From 0.22.0 to 0.23.0, the /v1/audio/transcriptions and /v1/audio/translations routes call request.file.read() to fully materialize an uploaded audio file into memory before vLLM checks the documented VLLM_MAX_AUDIO_CLIP_FILESIZE_MB compressed upload size limit (default 25 MB) later in the speech-to-text preprocessing step, so an API caller who can reach those routes can submit an oversized multipart upload and cause vLLM to allocate memory proportional to the uploaded file size before the request is rejected as too large, creating memory pressure or terminating the process depending on deployment resource limits. 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). 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). 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). 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). 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 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 is a high-throughput and memory-efficient inference and serving engine for LLMs. Prior to 0.24.0, a frontend-legal multi-request speculative decoding workload can cause the rejection sampler to produce a recovered token equal to the model vocabulary size boundary value, which is then converted to negative one when the engine selects the next live token for a request and is written back into the drafter's input ids; that out-of-vocabulary value is later consumed by the model's embedding and attention path and crashes the engine worker with a GPU device-side assertion. The same triggering request sequence is reachable through the public gRPC Generate and Abort endpoints, so a remote client that can send generation requests can crash the shared engine worker, aborting concurrent requests and causing a service-wide denial of service for other clients of the deployment until the worker is restarted. This issue is fixed in version 0.24.0.

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.

In etcd before versions 3.3.23 and 3.4.10, a large slice causes panic in decodeRecord method. The size of a record is stored in the length field of a WAL file and no additional validation is done on this data. Therefore, it is possible to forge an extremely large frame size that can unintentionally panic at the expense of any RAFT participant trying to decode the WAL.

HAX CMS NodeJs allows users to manage their microsite universe with a NodeJs backend. In versions 11.0.8 and below, the HAX CMS NodeJS application crashes when an authenticated attacker provides an API request lacking required URL parameters. This vulnerability affects the listFiles and saveFiles endpoints. This vulnerability exists because the application does not properly handle exceptions which occur as a result of changes to user-modifiable URL parameters. This is fixed in version 11.0.9.

Improper input validation in Windows Local Security Authority Subsystem Service (LSASS) allows an authorized attacker to deny service over a network.

Laminas Diactoros provides PSR HTTP Message implementations. In versions 2.18.0 and prior, 2.19.0, 2.20.0, 2.21.0, 2.22.0, 2.23.0, 2.24.0, and 2.25.0, users who create HTTP requests or responses using laminas/laminas-diactoros, when providing a newline at the start or end of a header key or value, can cause an invalid message. This can lead to denial of service vectors or application errors. The problem has been patched in following versions 2.18.1, 2.19.1, 2.20.1, 2.21.1, 2.22.1, 2.23.1, 2.24.1, and 2.25.1. As a workaround, validate HTTP header keys and/or values, and if using user-supplied values, filter them to strip off leading or trailing newline characters before calling `withHeader()`.

In a CVX cluster, an EOS switch connected to a CVX server is not resilient to certain malformed messages received from the connected CVX server. Similarly, the CVX server is not resilient to certain malformed messages received from the connected EOS switch. This leads to either a Sysdb agent crash on the EOS device causing a soft reset of the switch or agent crashes on the CVX server causing instability of the CVX cluster. An attacker could use this behavior to create a denial of service (DoS) scenario. Note that this would require the attacker to already have a high privilege access to the connected device to be able to send custom TCP packets. EOS switches that are not connected to a CVX server are not impacted.

IBM Watson Knowledge Catalog on Cloud Pak for Data 4.0 could allow an authenticated user send a specially crafted request that could cause a denial of service. IBM X-Force ID: 251704.

An improper input validation vulnerability in the TLS certificate generation function allows an attacker to cause a Denial-of-Service (DoS) condition which can only be reverted via a factory reset. This issue affects: Lanner Inc IAC-AST2500A standard firmware version 1.10.0.

Nessus versions 8.6.0 and earlier were found to contain a Denial of Service vulnerability due to improper validation of specific imported scan types. An authenticated, remote attacker could potentially exploit this vulnerability to cause a Nessus scanner to become temporarily unresponsive.

IBM Security Verify Access 10.0.0, 10.0.1, 10.0.2, 10.0.3, 10.0.4, and 10.0.5 could allow an attacker to crash the webseald process using specially crafted HTTP requests resulting in loss of access to the system. IBM X-Force ID: 247635.

TensorFlow is an Open Source Machine Learning Framework. In versions prior to 2.11.1 a malicious invalid input crashes a tensorflow model (Check Failed) and can be used to trigger a denial of service attack. A proof of concept can be constructed with the `Convolution3DTranspose` function. This Convolution3DTranspose layer is a very common API in modern neural networks. The ML models containing such vulnerable components could be deployed in ML applications or as cloud services. This failure could be potentially used to trigger a denial of service attack on ML cloud services. An attacker must have privilege to provide input to a `Convolution3DTranspose` call. This issue has been patched and users are advised to upgrade to version 2.11.1. There are no known workarounds for this vulnerability.

Windows CryptoAPI Denial of Service Vulnerability

Apache Airflow, versions before 2.6.3, is affected by a vulnerability that allows an attacker to cause a service disruption by manipulating the run_id parameter. This vulnerability is considered low since it requires an authenticated user to exploit it. It is recommended to upgrade to a version that is not affected

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

IBM Db2 for Linux, UNIX and Windows (includes DB2 Connect Server) 11.5 CLI is vulnerable to a denial of service when a specially crafted request is used. IBM X-Force ID: 268073.

In Octopus Deploy before 2019.10.6, an authenticated user with TeamEdit permission could send a malformed Team API request that bypasses input validation and causes an application level denial of service condition. (The fix for this was also backported to LTS 2019.9.8 and LTS 2019.6.14.)

Improper syscall input validation in the ASP Bootloader may allow a privileged attacker to read memory out-of-bounds, potentially leading to a denial-of-service.

Insufficient syscall input validation in the ASP Bootloader may allow a privileged attacker to read memory outside the bounds of a mapped register potentially leading to a denial of service.

A Denial of service (DoS) vulnerability in FortiClient for Linux 6.2.1 and below may allow an user with low privilege to cause FortiClient processes running under root privilege crashes via sending specially crafted IPC client requests to the fctsched process due the nanomsg not been correctly validated.

A vulnerability in the web interface of Cisco Wireless LAN Controller Software could allow a low-privileged, authenticated, remote attacker to cause a denial of service (DoS) condition on an affected device. The vulnerability exists due to a failure of the HTTP parsing engine to handle specially crafted URLs. An attacker could exploit this vulnerability by authenticating with low privileges to an affected controller and submitting the crafted URL to the web interface of the affected device. Conversely, an unauthenticated attacker could exploit this vulnerability by persuading a user of the web interface to click the crafted URL. A successful exploit could allow the attacker to cause an unexpected restart of the device, resulting in a DoS condition.

A vulnerability in the implementation of the Intermediate System&ndash;to&ndash;Intermediate System (IS&ndash;IS) routing protocol functionality in Cisco IOS XR Software could allow an authenticated, remote attacker to cause a denial of service (DoS) condition in the IS&ndash;IS process. The vulnerability is due to improper handling of a Simple Network Management Protocol (SNMP) request for specific Object Identifiers (OIDs) by the IS&ndash;IS process. An attacker could exploit this vulnerability by sending a crafted SNMP request to the affected device. A successful exploit could allow the attacker to cause a DoS condition in the IS&ndash;IS process.

OSIsoft PI SQL Data Access Server (aka OLE DB) 2016 1.5 allows remote authenticated users to cause a denial of service (service outage and data loss) via a message.

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.

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.

An authenticated, remote attacker may use a improper input validation vulnerability in the CmpApp/CmpAppBP/CmpAppForce Components of multiple CODESYS products in multiple versions to read from an invalid address which can lead to a denial-of-service condition.

IBM Db2 for Linux, UNIX and Windows (includes DB2 Connect Server) 11.5 under certain circumstances could allow an authenticated user to the database to cause a denial of service when a statement is run on columnar tables. IBM X-Force ID: 273393.

wire-ios is the iOS version of Wire, an open-source secure messaging app. In wire-ios versions 3.8.0 and prior, a vulnerability exists that can cause a denial of service between users. If a user has an invalid assetID for their profile picture and it contains the " character, it will cause the iOS client to crash. The vulnerability is patched in wire-ios version 3.8.1.

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.

An input validation vulnerability exists in the Rockwell Automation Sequence Manager™ which could allow a malicious user to send malformed packets to the server and cause a denial-of-service condition. If exploited, the device would become unresponsive, and a manual restart will be required for recovery. Additionally, if exploited, there could be a loss of view for the downstream equipment sequences in the controller. Users would not be able to view the status or command the equipment sequences, however the equipment sequence would continue to execute uninterrupted.

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

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

The decodenetnum function in ntpd in NTP 4.2.x before 4.2.8p4, and 4.3.x before 4.3.77 allows remote attackers to cause a denial of service (assertion failure) via a 6 or mode 7 packet containing a long data value.