vllm-project vLLM

• CVE-2026-115918 Medium 6.5 vLLM: Responses API bypasses trust_request_chat_template for chat_template_kwargs 0.4%
• CVE-2026-115917 Low 3.4 vLLM: Permissive default CORS origins allow cross-origin requests to the vLLM HTTP server 0.2%
• CVE-2026-115916 Medium 6.5 vLLM: Out-of-vocabulary prompt token id passes input validation and triggers a CUDA device-side assert (engine crash / DoS) 0.3%
• CVE-2026-115915 High 8.1 vLLM: Untrusted EPS inputs can trigger denial of service or Ghostscript RCE 0.7%
• CVE-2026-115914 Medium 6.5 vLLM DeepSeek-OCR image processing vulnerable to denial of service 0.4%
• CVE-2026-115913 Medium 6.5 vLLM: Phi-4 multimodal audio encoder materializes and retains full attention matrices: a single max-length audio OOM-crashes the vLLM engine 0.3%
• CVE-2026-115912 High 7.5 vLLM: Public multimodal SHM handle forgery reaches worker-side `pickle.loads()` with insecure serialization disabled 0.5%
• CVE-2026-115911 Medium 6.5 vLLM: DoS via unbounded chat_template Jinja2 rendering exhausts the renderer thread pool 0.3%
• CVE-2026-116181 Medium 6.5 vLLM 0.20.1 before 0.29.0 DoS and GPU Out-of-Bounds Write via stop_token_ids 0.4%
• CVE-2026-116180 High 7.7 vLLM from 0.8.5 before 0.31.0 Cross-Tenant DoS via Uninitialized LogprobsTensors 0.3%
• CVE-2026-116179 High 7.5 vLLM 0.18.0 before 0.31.0 Memory Exhaustion via media_io_kwargs num_frames Override 0.3%
• CVE-2026-116178 Medium 5.9 vLLM 0.20.0 before 0.30.0 DoS via /inference/v1/generate mm_placeholders 0.3%
• CVE-2026-116177 High 7.5 vLLM 0.10.2 before 0.25.0 DoS via Nemotron-VL Processor Decompression-Bomb Guard Bypass 0.3%
• CVE-2026-116176 High 8.2 vLLM 0.9.2 before 0.24.0 SSRF via X-Request-Id in P2pNcclConnector 0.3%
• CVE-2026-116175 Medium 6.5 vLLM 0.19.0 before 0.24.0 Audio Decompression Bomb DoS via Chat Completions 0.3%
• CVE-2026-116174 Medium 5.9 vLLM 0.22.0 before 0.24.0 Denial of Service via Rust gRPC Generate 0.3%
• CVE-2026-116173 Medium 5.9 vLLM 0.22.0 before 0.24.0 Denial of Service via gRPC output_candidates.token_ids 0.3%
• CVE-2026-116172 Medium 5.9 vLLM 0.22.0 before 0.24.0 DoS via Rust gRPC CandidateTokens.top_n 0.3%
• CVE-2026-114055 High 7.5 vLLM 0.7.3 before 0.31.0 DoS via Unvalidated mm_processor_kwargs 0.4%
• CVE-2026-114054 Medium 6.5 vLLM before 0.31.0 Cross-Request Cache Aliasing via EXIF ImageID and Prefix-Cache Keys 0.3%
• CVE-2026-114053 High 8.1 vLLM 0.7.3 before 0.31.0 RCE via mm_processor_kwargs.code_revision 0.5%
• CVE-2026-114052 High 7.5 vLLM 0.25.0 before 0.31.0 MOSS-Audio Memory Exhaustion via /tokenize mm_processor_kwargs 0.5%
• CVE-2026-114051 High 7.5 vLLM before 0.31.0 Denial of Service via min_tokens in /inference/v1/generate 0.4%
• CVE-2026-114050 High 7.5 vLLM before 0.31.0 Remote Engine Crash via Forged Multimodal UUID 0.2%
• CVE-2026-114049 Low 3.7 vLLM before 0.31.0 Stale allowed_token_ids Mask via InputBatch.condense 0.3%
• CVE-2026-109311 Medium 6.5 vLLM before 0.29.0 Heap Overflow DoS via IAMF Audio Transcription Upload 0.5%
• CVE-2026-105922 Medium 4.3 vllm-project vLLM Penalty utils.py get_token_bin_counts_and_mask denial of service 0.3%
• CVE-2026-105775 Medium 4.3 vllm-project vLLM Completions Request mamba_mixer2.py conv_ssm_forward out-of-bounds 0.3%
• CVE-2026-105760 Medium 5.3 vLLM: GLMGA video sampling permits request-driven CPU and memory exhaustion 0.3%
• CVE-2026-105759 Medium 5.9 vLLM: Unbounded Prometheus label cardinality from attacker-controlled HTTP method tokens in the vLLM Rust frontend metrics middleware (unauthenticated denial of service) 0.4%
• CVE-2026-105758 Medium 5.3 vLLM: Qwen2-VL / Qwen3-VL video samplers bound on request-controlled max_frames, which the num_frames ceiling does not reach 0.3%
• CVE-2026-105757 Medium 6.5 vLLM: Structured-output request errors escape the request boundary and terminate the shared EngineCore — engine-fatal denial of service (3 sites) 0.3%
• CVE-2026-105756 Medium 6.5 vLLM: Loose `cache_salt` validation lets a single request kill EngineCore on LMCache-MP deployments — uncaught downstream `ValueError` denial of service 0.3%
• CVE-2026-105755 Medium 4.2 vLLM: Flash late-interaction scoring caches query embeddings under a caller-controlled request id — cross-request integrity break and induced errors on `/score` and `/rerank` 0.2%
• CVE-2026-105754 Medium 6.5 vLLM: Scale-out disaggregated multimodal transport trusts caller-supplied features 0.3%
• CVE-2026-105753 Medium 6.5 vLLM: Mirrored multimodal IPC caches desync after a rejected request — a later request reusing the same media hash trips a receiver assertion in the engine core 0.4%
• CVE-2026-105752 Low 3.1 vLLM: Harmony tool continuations drop `cache_salt` — restoring a cross-tenant prefix-cache membership oracle 0.2%
• CVE-2026-103241 Medium 5.3 vllm-project vLLM Gemma4UnifiedParser gemma4.rs denial of service 0.7%
• CVE-2026-100654 Medium 6.5 vLLM before 0.29.0 Denial of Service via out-of-range stop_token_ids 0.3%
• CVE-2026-100653 Medium 6.5 vLLM 0.22.1 before 0.28.0 Incomplete Artifact Pin Propagation 0.3%
• CVE-2026-100652 Medium 5.9 vLLM 0.22.0 through 0.23.0 Denial of Service via stop_token_ids 0.4%
• CVE-2026-100651 Medium 6.5 vllm before 0.29.0 Denial of Service via Decoder Prompt Length Bypass 0.4%
• CVE-2026-100650 Medium 6.5 vLLM before 0.29.0 Resource Exhaustion via Unbounded Media Materialization 0.8%
• CVE-2026-100649 Low 3.7 vLLM before 0.29.0 Resource Limit Bypass via Sampler Subclass 0.3%
• CVE-2026-100648 Medium 5.3 vllm before 0.29.0 Uncontrolled Resource Consumption via Audio Decoding 0.4%
• CVE-2026-100647 Medium 5.3 vLLM before 0.29.0 CPU Exhaustion via unbounded cache_salt 0.3%
• CVE-2026-94627 High 7.5 vLLM through 0.29.0 GPU KV Cache Leak via Mooncake Transfer ID Collision 0.6%
• CVE-2026-94626 High 7.5 vLLM through 0.29.0 Memory Exhaustion via Unvalidated NIXL tp_size 0.6%
• CVE-2026-94625 Medium 5.3 vLLM through 0.29.0 Resource Exhaustion via Ownerless Mooncake Transfer Placeholders 0.5%
• CVE-2026-94624 High 7.5 vLLM through 0.29.0 Denial of Service via Unbounded P2P KV Offloading Sessions 0.4%