vllm-project vLLM

• CVE-2026-94623 High 7.5 vLLM through 0.29.0 Denial of Service via NIXL Multi-Prompt Assertion Failure 0.6%
• CVE-2026-94622 High 7.5 vLLM through 0.29.0 Denial of Service via Incomplete NIXL KV Transfer Metadata 0.6%
• CVE-2026-93989 Low 3.1 vLLM through 0.29.0 Cross-Request Logits Corruption via bad_words 0.2%
• CVE-2026-93841 Low 3.7 vLLM through 0.29.0 Adjacent-Request Sampler State Corruption via Unvalidated Prompt Token IDs 0.2%
• CVE-2026-93840 Low 3.7 vLLM before 0.29.0 Cross-Request Logits Corruption via allowed_token_ids 0.2%
• CVE-2026-93592 High 7.5 vLLM before 0.28.0 Denial of Service via negative token ID 0.5%
• CVE-2026-93436 High 7.5 vLLM through 0.29.0 Memory Exhaustion via Rejected Requests 0.8%
• CVE-2026-69147 Medium 6.5 vLLM: Request-selected PyNvVideoCodec GPU decode bypasses static VRAM reservation 0.6%
• CVE-2026-57173 Medium 6.5 vLLM: Unauthenticated audio decompression-bomb DoS in /v1/chat/completions 0.7%
• CVE-2026-92365 Medium 4.3 vllm-project vllm thinking_budget_state.py algorithmic complexity 0.5%
• CVE-2026-92220 Medium 5.3 vllm-project vLLM MoRIIO Acknowledgement moriio_connector.py MoRIIOWrapper._handle_release_message resource consumption 0.7%
• CVE-2026-90878 Medium 4.3 vllm-project vLLM Jinja Template Rendering completions resource consumption 0.5%
• CVE-2026-90713 Low 3.3 vllm-project vLLM tiktoken vocab File mod.rs new denial of service 0.2%
• CVE-2026-90555 Medium 6.5 vLLM before 0.28.0 Denial of Service via Audio Header 0.5%
• CVE-2026-90554 Medium 6.2 vLLM before 0.28.0 Denial of Service via audio extraction 0.2%
• CVE-2026-90553 High 7.8 vLLM before 0.28.0 Remote Code Execution via LlavaOnevision2 processor 0.3%
• CVE-2026-78684 Medium 5.3 vLLM before 0.27.0 Denial of Service via DeepStream Backend 0.6%
• CVE-2026-73560 Medium 6.5 vLLM: SSRF + arbitrary local file read in MiMoV2OmniMultiModalProcessor `_fetch_image` and audio loader bypass MediaConnector protections 0.4%
• CVE-2026-71486 Medium 4.3 vLLM: Derender endpoints decode caller-supplied GenerateResponse token IDs without output bounds 0.4%
• CVE-2026-73559 Medium 6.5 vLLM: Completion prompt lists fan out into unbounded engine requests 0.6%
• CVE-2026-73558 Medium 5.3 vLLM: Cross-User Data Leak Vulnerability 0.4%
• CVE-2026-73557 Medium 6.3 vLLM: Incomplete CVE-2025-62164 remediation can be bypassed by concurrent prompt parts 0.2%
• CVE-2026-73556 Medium 5.3 vLLM: ReDoS via structured_outputs.regex in the lm-format-enforcer backend (no compile timeout) — missed sibling of CVE-2026-55574 0.5%
• CVE-2026-73555 Medium 5.3 vLLM: Unauthenticated Internal Path and Username Disclosure via Validation Error Messages 0.4%
• CVE-2026-55514 High 7.1 vLLM denial of service via prompt embeds on M-RoPE models 0.7%
• CVE-2026-55574 High 8.7 vLLM: ReDoS via structured_outputs.regex compiled without timeout in xgrammar and outlines backends 0.6%
• CVE-2026-54234 High 7.5 vLLM: Remote DoS in vLLM via Invalid Recovered Token Reinjection 0.6%
• CVE-2026-55646 Medium 6.5 vLLM speech-to-text endpoints allocate full upload before enforcing the audio file-size limit 0.5%
• CVE-2026-47155 Medium 6.5 vLLM: Artifact Pin Decay in vLLM allows pinned deployments to load unpinned code, weights, and processors 0.2%
• CVE-2026-41523 High 7.5 vLLM: Security Check Bypass via assert Statement in Activation Function Loading Allows Arbitrary Code Execution 0.9%
• CVE-2026-54232 High 8.8 vLLM: Dependency Confusion Vulnerability in vLLM Dockerfile 0.6%
• CVE-2026-54233 Medium 6.5 vLLM: OOM Denial of Service via Audio Decompression Bomb 0.4%
• CVE-2026-54236 Medium 5.3 vLLM: incomplete CVE-2026-22778 fix leaks PIL repr addresses via Anthropic router 0.9%
• CVE-2026-54235 Medium 6.9 vLLM: temperature=NaN and temperature=Infinity bypass validation and propagate to GPU kernels 0.4%
• CVE-2026-48746 Critical 9.1 vLLM: OpenAI auth bypass 1.2%
• CVE-2026-53923 Medium 5.3 vLLM GGUF Kernels: int64_t to int truncation of tensor dimensions causes GPU buffer overflow 0.5%
• CVE-2026-12491 Medium 4.8 Vllm: vllm: image exif rotation & png trns transparency not normalized, causing mismatch between model input and expectations 0.2%
• CVE-2026-9540 Medium 5.3 vllm-project vllm OpenAI-compatible Serving Path denial of service 0.4%
• CVE-2026-44223 Medium 6.5 vLLM: extract_hidden_states speculative decoding crashes server on any request with penalty parameters 0.4%
• CVE-2026-44222 Medium 6.5 vLLM: Remote DoS via Special-Token Placeholders 0.4%
• CVE-2026-34756 Medium 6.5 vLLM Affected by Unauthenticated OOM Denial of Service via Unbounded `n` Parameter in OpenAI API Server 0.4%
• CVE-2026-34755 Medium 6.5 vLLM Affected by Denial of Service via Unbounded Frame Count in video/jpeg Base64 Processing 0.5%
• CVE-2026-34753 Medium 5.4 vLLM affected by Server-Side Request Forgery (SSRF) in `download_bytes_from_url ` 0.3%
• CVE-2026-34760 Medium 5.9 vLLM: Downmix Implementation Differences as Attack Vectors Against Audio AI Models 0.3%
• CVE-2026-27893 High 8.8 vLLM's hardcoded trust_remote_code=True in NemotronVL and KimiK25 bypasses user security opt-out 1.4%
• CVE-2026-25960 High 7.1 SSRF Protection Bypass in vLLM 0.6%
• CVE-2026-22778 Critical 9.8 vLLM leaks a heap address when PIL throws an error 18.3%
• CVE-2026-24779 High 7.1 vLLM vulnerable to Server-Side Request Forgery (SSRF) in `MediaConnector` 0.6%
• CVE-2026-22807 High 8.8 vLLM affected by RCE via auto_map dynamic module loading during model initialization 1.0%
• CVE-2026-22773 Medium 6.5 vLLM is vulnerable to DoS in Idefics3 vision models via image payload with ambiguous dimensions 0.4%