Vllm vulnerabilities and fixes
Exploited and high-severity Vllm vulnerabilities tracked by VulnFeed, with CVSS and EPSS scores, affected versions and fixes.
- 8.1CVE-2026-115915vLLM: Untrusted EPS inputs can trigger denial of service or Ghostscript RCEFix availableVllm Vllm
1 min read - 7.5CVE-2026-115912vLLM: Public multimodal SHM handle forgery reaches worker-side `pickle.loads()` with insecure serialization disabledRCEFix availableVllm Vllm
1 min read - 8.2CVE-2026-116176vLLM from 0.9.2 before 0.24.0 contains a server-side request forgery vulnerability that allows API callers to force outbound connections by embedding Fix availableVllm Vllm
1 min read - 8.1CVE-2026-114053vLLM from 0.7.3 before 0.31.0 forwards the request-supplied mm_processor_kwargs.code_revision to Transformers AutoProcessor.from_pretrained, letting cRCEFix availableVllm Vllm
1 min read - 7.5CVE-2026-114055vLLM from 0.7.3 before 0.31.0 contains an uncontrolled resource consumption vulnerability that allows remote clients to control multimodal processor pFix availableVllm Vllm
1 min read - 7.5CVE-2026-114050vLLM before 0.31.0 uses a client-supplied multimodal uuid verbatim as cache identity with no consistency check on cache hits, allowing engine crashes.Fix availableVllm Vllm
1 min read - 7.5CVE-2026-114051vLLM before 0.31.0 skips the min_tokens <= max_tokens check for /inference/v1/generate requests with a null max_tokens, so the EngineCore input threadFix availableVllm Vllm
1 min read - 7.5CVE-2026-114052vLLM after 0.24.0 and before 0.31.0 contains a memory exhaustion vulnerability in MOSS-Audio model serving that caches a processor for every distinct Fix availableVllm Vllm
1 min read