CVE-2026-54234 vLLM: Remote DoS in vLLM via Invalid Recovered Token Reinjection

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 convert...

CVE-2026-55646

CVE-2026-55646 (vLLM) affects vLLM versions 0.22.0–0.23.0. The routes /v1/audio/transcriptions and /v1/audio/translations call request.file.read() to fully materialize an uploaded audio file before enforcing the documented size limit (default 25 MB). This can cause the server to allocate memory p...

CVE-2026-55646 vLLM speech-to-text endpoints allocate full upload before enforcing the audio file-size limit

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 VLLMMAXAUDIOCLIPFILESIZEMB...

CVE-2026-55646 vLLM speech-to-text endpoints allocate full upload before enforcing the audio file-size limit

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 VLLMMAXAUDIOCLIPFILESIZEMB...

PT-2026-56000

Name of the Vulnerable Software and Affected Versions vLLM versions prior to 0.24.0 Description The structured outputs.regex API parameter allows user-supplied regular expression strings to be passed to grammar compiler backends without a compilation timeout. In the xgrammar backend, the string i...

PT-2026-55999

Name of the Vulnerable Software and Affected Versions vLLM versions 0.12.0 through 0.23.x Description A flaw in the EngineCore of the library occurs when a remote authorized user sends a pure prompt embeds payload to the '/v1/completions' endpoint using a model that implements M-RoPE Multimodal...

GHSA-29PF-2H5F-8G72 vulnerabilities

Vulnerabilities for packages: tritonserver-backend-vllm-cuda-12.9, text-generation-inference, nemo...

CVE-2026-4372 vulnerabilities

Vulnerabilities for packages: tritonserver-backend-vllm-cuda-12.9, text-generation-inference, nemo...

GHSA-6PR9-RP53-2PMC vulnerabilities

Vulnerabilities for packages: vllm-cuda-13.2, vllm-openai-cuda-13.0...

CVE-2026-54235 vulnerabilities

Vulnerabilities for packages: vllm-openai-cuda-13.0, vllm-cuda-13.2, py3-vllm-cuda-12.4...

GHSA-7H4P-RFFG-7823 vulnerabilities

Vulnerabilities for packages: vllm-openai-cuda-13.0, vllm-cuda-13.2, py3-vllm-cuda-12.4...

GHSA-5JV2-G5WQ-CMR4 vulnerabilities

Vulnerabilities for packages: vllm-cuda-13.2, vllm-openai-cuda-13.0...

CVE-2026-54233 vulnerabilities

Vulnerabilities for packages: vllm-cuda-13.2, vllm-openai-cuda-13.0...

CVE-2026-53923 vulnerabilities

Vulnerabilities for packages: vllm-cuda-13.2, vllm-openai-cuda-13.0...

CVE-2026-54236 vulnerabilities

Vulnerabilities for packages: vllm-cuda-13.2, vllm-openai-cuda-13.0...

CVE-2026-12491 vulnerabilities

Vulnerabilities for packages: vllm-cuda-13.2, vllm-openai-cuda-13.0...

GHSA-HGG8-FQQC-VFMW vulnerabilities

Vulnerabilities for packages: vllm-cuda-13.2, vllm-openai-cuda-13.0...

GHSA-8JR5-V98P-W75M vulnerabilities

Vulnerabilities for packages: vllm-cuda-13.2, vllm-openai-cuda-13.0...

PYSEC-2026-566 vLLM Deserialization of Untrusted Data vulnerability

vllm-project vllm version v0.6.2 contains a vulnerability in the MessageQueue.dequeue API function. The function uses pickle.loads to parse received sockets directly, leading to a remote code execution vulnerability. An attacker can exploit this by sending a malicious payload to the MessageQueue,...

PYSEC-2026-567 vLLM Allows Remote Code Execution via PyNcclPipe Communication Service

Impacted Environments This issue ONLY impacts environments using the PyNcclPipe KV cache transfer integration with the V0 engine. No other configurations are affected. Summary vLLM supports the use of the PyNcclPipe class to establish a peer-to-peer communication domain for data transmission...