Skip to content
CVE
by OpenHack
Browse
All CVEs
Every CVE, newest first, with filters
CVEs by company
Microsoft, Cisco, Fortinet and 840 more
CVEs by package
Open-source packages on npm, PyPI, Maven, Go…
Critical
About
Search CVEs, products, packages and vendors
/
Packages
PyPI
vllm
vllm CVEs
PyPI package
Search CVEs, products, packages and vendors
Severity
Any
Published
All time
Exploited only
Version
113 CVEs
·
5 without a fix
·
CVE
Record
Severity
Status
Published
CVE-2026-115917
vLLM data exposure via wildcard CORS in the OpenAI-compatible server
Fixed in 0.32.0
Low
3.4
Fix available
Oct 28
CVE-2026-115913
vLLM denial of service in Phi-4 audio attention
Fixed in 0.32.0
Medium
6.5
Fix available
Oct 28
CVE-2026-115914
vLLM denial of service in DeepSeek OCR image handling
Fixed in 0.32.0
Medium
6.5
Fix available
Oct 28
CVE-2026-115915
vLLM command execution via EPS image decoding on affected hosts
Fixed in 0.32.0
High
8.1
Fix available
Oct 28
CVE-2026-115912
vLLM code execution in multimodal shared-memory caching
Fixed in 0.32.0
High
7.5
Fix available
Oct 28
CVE-2026-115916
vLLM denial of service in pre-tokenized prompt validation
Fixed in 0.32.0
Medium
6.5
Fix available
Oct 28
CVE-2026-115918
vLLM denial of service via chat template policy bypass in /v1/responses
Fixed in 0.32.0
Medium
6.5
Fix available
Oct 28
CVE-2026-115911
vLLM denial of service in chat template rendering
Fixed in 0.32.0
Medium
6.5
Fix available
Oct 28
CVE-2026-114053
vLLM code execution in multimodal processor revision selection
Fixed in 0.31.0
High
8.1
Fix available
Oct 27
CVE-2026-94624
vllm p2p never connected zmq_exhaustion
Fixed in 0.32.0
Medium
6.5
Fix available
Oct 27
CVE-2026-116178
vLLM denial of service in disaggregated multimodal generation
Fixed in 0.30.0
High
7.5
Fix available
Oct 25
CVE-2026-116177
vLLM denial of service via Nemotron-VL image decompression bomb
Fixed in 0.25.0
High
7.5
Fix available
Oct 25
CVE-2026-116176
vLLM data exposure via X-Request-Id connector address
Fixed in 0.24.0
High
8.2
Fix available
Oct 25
CVE-2026-116173
vLLM denial of service via gRPC logprob token IDs
Fixed in 0.24.0
High
7.5
Fix available
Oct 25
CVE-2026-114055
vLLM denial of service in multimodal processor parameters
Fixed in 0.31.0
High
7.5
Fix available
Oct 23
CVE-2026-114050
vLLM denial of service in multimodal UUID cache handling
Fixed in 0.31.0
Medium
6.5
Fix available
Oct 23
CVE-2026-114054
vLLM cached media exposure via shared cache key collisions
Fixed in 0.31.0
Unrated
Fix available
Oct 21
CVE-2026-114052
vLLM denial of service in MOSS-Audio processor cache
Fixed in 0.31.0
Unrated
Fix available
Oct 21
CVE-2026-105922
A security flaw has been discovered in vllm-project vLLM up to 0.31.0. This impacts the function get_token_bin_counts_and_mask of the file vllm/model_executor/layers/utils.py of the component Penalty
Fixed in 0.32.0
Unrated
Fix available
Oct 6
CVE-2026-105753
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
Fixed in 0.28.0
Medium
6.5
Fix available
Oct 6
CVE-2026-105752
vLLM: Harmony tool continuations drop `cache_salt` — restoring a cross-tenant prefix-cache membership oracle
Fixed in 0.30.0
Low
3.1
Fix available
Oct 6
CVE-2026-105758
vLLM: Qwen2-VL / Qwen3-VL video samplers bound on request-controlled max_frames, which the num_frames ceiling does not reach
Fixed in 0.30.0
Medium
5.3
Fix available
Oct 5
CVE-2026-105760
vLLM: GLMGA video sampling permits request-driven CPU and memory exhaustion
Fixed in 0.30.0
Medium
5.3
Fix available
Oct 5
CVE-2026-105754
vLLM: Scale-out disaggregated multimodal transport trusts caller-supplied features
Fixed in 0.30.0
Medium
6.5
Fix available
Oct 5
CVE-2026-105757
vLLM: Structured-output request errors escape the request boundary and terminate the shared EngineCore — engine-fatal denial of service (3 sites)
Fixed in 0.30.0
Medium
6.5
Fix available
Oct 5
CVE-2026-105755
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`
Fixed in 0.30.0
Medium
4.2
Fix available
Oct 5
CVE-2026-105756
vLLM: Loose `cache_salt` validation lets a single request kill EngineCore on LMCache-MP deployments — uncaught downstream `ValueError` denial of service
Fixed in 0.30.0
Medium
6.5
Fix available
Oct 5
CVE-2026-105759
vLLM is an inference and serving engine for large language models. Prior to 0.30.0, the Rust frontend's track_http_metrics middleware records the raw HTTP method token as a Prometheus label for reques
Fixed in 0.30.0
High
7.5
Fix available
Oct 5
CVE-2026-103241
A flaw has been found in vllm-project vLLM up to 0.26.0. This vulnerability affects unknown code of the file rust/src/parser/src/unified/gemma4.rs of the component Gemma4UnifiedParser. Executing a man
Fixed in 0.27.0
High
7.5
Fix available
Sep 30
CVE-2026-100653
vLLM is an inference and serving engine for large language models. In versions from 0.22.1 through 0.28.0, the operator-supplied model revision pin (--revision / --code-revision) is not propagated to
Fixed in 0.28.0
Unrated
Fix available
Sep 26
Showing
1–30
of 113
Previous
1
2
3
4
Next