CVE-2026-100652: Improper Input Validation

vLLM versions 0.22.0 through 0.23.0 fail to validate stoptokenids against vocabulary bounds in Rust HTTP and gRPC frontends, allowing out-of-vocabulary token IDs to reach MinTokensLogitsProcessor. Attackers can submit requests with mintokens greater than zero and out-of-vocabulary stoptokenids to...

PT-2026-99325

Name of the Vulnerable Software and Affected Versions vLLM versions prior to 0.29.0 Description The software fails to validate if the values provided in stop token ids are within the model vocabulary or logits range, checking only that they are integers. When min tokens is greater than 0, these I...

Important: Red Hat Security Advisory: Red Hat AI Base Images 3.5.1 (CUDA 13.0)

Red Hat AI Base Images 3.5.1 CUDA 13.0 is now available. Red Hat® AI Base Images...

Important: Red Hat Security Advisory: Red Hat AI Inference 3.4.5 (cuda)

Red Hat AI Inference 3.4.5 cuda is now available. Red Hat AI Inference...

GHSA-7M6H-X95X-82Q5 vulnerabilities

Vulnerabilities for packages: tritonserver-backend-vllm-cuda-13.0...

CVE-2026-73558 vulnerabilities

Vulnerabilities for packages: tritonserver-backend-vllm-cuda-13.0...

CVE-2026-93592

A flaw was found in vLLM. An unauthenticated attacker can exploit an input validation vulnerability by submitting negative token IDs to the /v1/embeddings and /pooling endpoints. This action triggers a CUDA device-side assertion, which poisons the GPU context and causes the engine to crash...

Important: Red Hat Security Advisory: Red Hat AI Base Images 3.4.2 (CUDA 13.0)

Red Hat AI Base Images 3.4.2 CUDA 13.0 is now available. Red Hat® AI Base Images...

Important: Red Hat Security Advisory: Red Hat AI Base Images 3.4.2 (CUDA 12.9)

Red Hat AI Base Images 3.4.2 CUDA 12.9 is now available. Red Hat® AI Base Images...

Important: Red Hat Security Advisory: Red Hat AI Inference Model Optimization Tools 3.4.5 (cuda)

Red Hat AI Inference Model Optimization Tools 3.4.5 cuda is now available. Red Hat AI Inference Model Optimization Tools...

SUSE CVE-2026-93592

vLLM versions before 0.28.0 fail to validate the lower bound of token IDs in the /v1/embeddings and /pooling endpoints, allowing unauthenticated attackers to crash the engine by submitting negative token IDs. A single request with a negative token ID triggers a CUDA device-side assertion that...

GHSA-8PW2-6JV3-MJ5J vulnerabilities

Vulnerabilities for packages: tritonserver-backend-vllm-cuda-13.0...

Improper Validation of Array Index

Overview vllm is an A high-throughput and memory-efficient inference and serving engine for LLMs Affected versions of this package are vulnerable to Improper Validation of Array Index via InputProcessor.validatemodelinput in the /v1/embeddings and /pooling endpoints, which validates only the uppe...

CVE-2026-69147 vulnerabilities

Vulnerabilities for packages: tritonserver-backend-vllm-cuda-13.0...

PYSEC-2026-3997

vLLM versions before 0.28.0 fail to validate the lower bound of token IDs in the /v1/embeddings and /pooling endpoints, allowing unauthenticated attackers to crash the engine by submitting negative token IDs. A single request with a negative token ID triggers a CUDA device-side assertion that...

PYSEC-0000-CVE-2026-93592

vLLM versions before 0.28.0 fail to validate the lower bound of token IDs in the /v1/embeddings and /pooling endpoints, allowing unauthenticated attackers to crash the engine by submitting negative token IDs. A single request with a negative token ID triggers a CUDA device-side assertion that...

PYSEC-2026-3997

vLLM versions before 0.28.0 fail to validate the lower bound of token IDs in the /v1/embeddings and /pooling endpoints, allowing unauthenticated attackers to crash the engine by submitting negative token IDs. A single request with a negative token ID triggers a CUDA device-side assertion that...

EUVD-2026-82911

vLLM versions before 0.28.0 fail to validate the lower bound of token IDs in the /v1/embeddings and /pooling endpoints, allowing unauthenticated attackers to crash the engine by submitting negative token IDs. A single request with a negative token ID triggers a CUDA device-side assertion that...

CVE-2026-93592

vLLM versions before 0.28.0 fail to validate the lower bound of token IDs in the /v1/embeddings and /pooling endpoints. An unauthenticated attacker can send a single request containing a negative token ID , triggering a CUDA device-side assertion that poisons the GPU context and causes all subseq...

SUSE CVE-2026-69147

vLLM is an inference and serving engine for large language models. Prior to 0.28.0, request bodies for Chat Completions and Responses can set mediaiokwargs.video.videobackend to pynvvideocodec, and MediaConnector.fetchvideo forwards that choice to VideoMediaIO even when startup configuration...