TI-2024-10229 CVE-2024-11041 · GHSA-5vqr-wprc-cpp7
Remote Code Execution in vllm-project/vllm
Priority HIGH 82
vllm-project vllm version v0.6.2 contains a vulnerability in the MessageQueue.dequeue() API function.
vllm-project
Records125All Threatist records
Exploited (KEV)0None in CISA KEV
CVSS critical11CVSS 9.0–10.0
CVSS high54CVSS 7.0–8.9
Exploited now
vllm-project records in CISA KEVNo vllm-project vulnerabilities are listed in CISA KEV.
Components
Products in the NVD configurations · select one to filter the recordsRecords
125 records
TI-2024-08404 CVE-2024-9053 · GHSA-cj47-qj6g-x7r4
Remote Code Execution in vllm-project/vllm
Priority HIGH 82
vllm-project vllm version 0.6.0 contains a vulnerability in the AsyncEngineRPCServer() RPC server entrypoints.
TI-2026-36017 CVE-2026-48746 · GHSA-94f4-hr76-p5j6
vLLM: OpenAI auth bypass
Priority HIGH 78
vLLM is an inference and serving engine for large language models (LLMs).
TI-2026-63056 CVE-2026-93592 · GHSA-c3vq-r33v-49cr
vLLM before 0.28.0 Denial of Service via negative token ID
Priority HIGH 77
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.
TI-2026-04308 CVE-2026-4944 · GHSA-g57c-wgqx-8wx7
Hardcoded trust_remote_code=True in vllm-project/vllm Bypasses User Security Control
Priority HIGH 73
vllm-project/vllm version 0.14.1 contains a vulnerability where the `trust_remote_code=True` parameter is hardcoded in two model implementation files (`vllm/model_executor/models/nemotron_vl.py` and `vllm/model_executor/…
TI-2026-39137 CVE-2026-54234 · GHSA-8wr5-jm2h-8r4f
vLLM: Remote DoS in vLLM via Invalid Recovered Token Reinjection
Priority HIGH 72
vLLM is a high-throughput and memory-efficient inference and serving engine for LLMs.
TI-2026-04816 CVE-2026-5497 · GHSA-wcwg-c5fc-9vrc
Unbounded Frame Count in video/jpeg Base64 Data URL Processing Leads to OOM DoS in vllm-project/vllm
Priority HIGH 72
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.
TI-2026-65616 CVE-2026-103241 · GHSA-328r-mpfv-qc55
vllm-project vLLM Gemma4UnifiedParser gemma4.rs denial of service
Priority HIGH 70
A flaw has been found in vllm-project vLLM up to 0.26.0.
TI-2026-08382 CVE-2026-9540 · GHSA-98f3-hwg4-4rf7
vllm-project vllm OpenAI-compatible Serving Path denial of service
Priority HIGH 70
A vulnerability was identified in vllm-project vllm 0.19.0.
TI-2026-64813 CVE-2026-100653 · GHSA-r3cf-2wgr-7xh9
vLLM 0.22.1 before 0.28.0 Incomplete Artifact Pin Propagation
Priority HIGH 69
vLLM is an inference and serving engine for large language models.
TI-2026-18296 CVE-2026-22778 · GHSA-4r2x-xpjr-7cvv
vLLM leaks a heap address when PIL throws an error
Priority HIGH 67
vLLM is an inference and serving engine for large language models (LLMs).
Priority HIGH 66
vLLM is an inference and serving engine for large language models.
TI-2025-29277 CVE-2025-46560 · GHSA-vc6m-hm49-g9qg
vLLM phi4mm: Quadratic Time Complexity in Input Token Processing leads to denial of service
Priority HIGH 66
vLLM is a high-throughput and memory-efficient inference and serving engine for LLMs.
TI-2026-71127 CVE-2026-109311 · GHSA-8wc5-pcgr-pg48
Heap overflow in vLLM audio transcription crashes the API process
Priority MEDIUM 64
vLLM before 0.29.0 passes uploaded audio to the PyAV/FFmpeg av.open() fallback with only a file-size check, exposing FFmpeg's IAMF parser heap overflow.
TI-2026-64811 CVE-2026-100651 · GHSA-g43f-v3gj-45x5
vllm before 0.29.0 Denial of Service via Decoder Prompt Length Bypass
Priority MEDIUM 64
vLLM before 0.29.0 fails to enforce decoder prompt-length validation on the disaggregated serving endpoint /inference/v1/generate.
TI-2026-64810 CVE-2026-100650 · GHSA-j8gm-f3cq-6c27
vLLM before 0.29.0 Resource Exhaustion via Unbounded Media Materialization
Priority MEDIUM 64
vLLM through 0.29.0 fetches and fully materializes remote or inline media before enforcing its documented media controls (the VLLM_MAX_AUDIO_CLIP_FILESIZE_MB compressed-audio size cap, default 25 MB, and the per-modality…
TI-2026-19980 CVE-2026-24779 · GHSA-qh4c-xf7m-gxfc
vLLM vulnerable to Server-Side Request Forgery (SSRF) in `MediaConnector`
Priority MEDIUM 64
vLLM is an inference and serving engine for large language models (LLMs).
TI-2026-51646 CVE-2026-73556 · GHSA-48jh-3gj7-fg8v
Regular expression denial of service in vllm-project vllm
Priority MEDIUM 63
vLLM is an inference and serving engine for large language models.
TI-2026-39139 CVE-2026-54236 · GHSA-hgg8-fqqc-vfmw
vLLM: incomplete CVE-2026-22778 fix leaks PIL repr addresses via Anthropic router
Priority MEDIUM 63
vLLM is an inference and serving engine for large language models (LLMs).
TI-2026-51650 CVE-2026-73560 · GHSA-4hhp-h66f-j5j7
Server-side request forgery in vllm-project vllm
Priority MEDIUM 62
vLLM is an inference and serving engine for large language models.
TI-2026-51649 CVE-2026-73559 · GHSA-87x5-vmc3-756j
vLLM: Completion prompt lists fan out into unbounded engine requests
Priority MEDIUM 62
vLLM is an inference and serving engine for large language models.
TI-2026-48528 CVE-2026-69147 · GHSA-8pw2-6jv3-mj5j
vLLM: Request-selected PyNvVideoCodec GPU decode bypasses static VRAM reservation
Priority MEDIUM 62
vLLM is an inference and serving engine for large language models.
TI-2026-40172 CVE-2026-55646 · GHSA-v82g-2437-67m2
vLLM speech-to-text endpoints allocate full upload before enforcing the audio file-size limit
Priority MEDIUM 62
vLLM is an inference and serving engine for large language models.
TI-2026-32202 CVE-2026-44223 · GHSA-83vm-p52w-f9pw
vLLM: extract_hidden_states speculative decoding crashes server on any request with penalty parameters
Priority MEDIUM 62
vLLM is an inference and serving engine for large language models (LLMs).
TI-2025-31082 CVE-2025-48944 · GHSA-vrq3-r879-7m65
vLLM Tool Schema allows DoS via Malformed pattern and type Fields
Priority MEDIUM 62
vLLM is an inference and serving engine for large language models (LLMs).