Vllm
by Vllm
Source repositories
CVEs (145)
| CVE | Vendor / Product | Sev | Risk | CVSS | EPSS | KEV | Published | Description |
|---|---|---|---|---|---|---|---|---|
| CVE-2026-115911 | Med | 0.35 | 6.5 | 0.00 | Oct 26, 2026 | vLLM is an inference and serving engine for large language models. From 0.19.0 until 0.32.0, authenticated callers can supply a request chat template when no server-side --chat-template is configured, or a malicious model repository can provide a template that is compiled and… | ||
| CVE-2026-116181 | Med | 0.35 | 6.5 | 0.00 | Oct 25, 2026 | vLLM from 0.20.1 before 0.29.0 contains an improper input validation vulnerability that allows remote attackers to crash the engine by supplying out-of-vocabulary stop_token_ids. Attackers can send OpenAI chat, completion, generate, or batch requests with invalid IDs to… | ||
| CVE-2026-116175 | Med | 0.35 | 6.5 | 0.00 | Oct 25, 2026 | vLLM from 0.19.0 before 0.24.0 contains a decompression bomb vulnerability because the AudioMediaIO multimodal path calls load_audio without the max_duration_s decoded-duration limit. Authenticated attackers can send small compressed audio via input_audio or audio_url in… | ||
| CVE-2026-114054 | Med | 0.35 | 6.5 | 0.00 | Oct 21, 2026 | vLLM before 0.31.0 contains a cache aliasing vulnerability that allows authenticated attackers to collide shared cache keys by supplying caller-controlled values such as EXIF ImageID or cache_salt. Attackers can set a UUID EXIF ImageID to receive another request's cached… | ||
| CVE-2026-109311 | Med | 0.35 | 6.5 | 0.00 | Oct 12, 2026 | 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. Authenticated remote attackers can repeatedly upload malformed IAMF containers to /v1/audio/transcriptions, corrupting the… | ||
| CVE-2026-105757 | Med | 0.35 | 6.5 | 0.00 | Oct 5, 2026 | vLLM is an inference and serving engine for large language models. Prior to 0.30.0, structured-output request failures can escape request-scoped validation and reach the EngineCore fatal-error path. A per-request backend mismatch can re-raise a grammar compilation exception,… | ||
| CVE-2026-105756 | Med | 0.35 | 6.5 | 0.00 | Oct 5, 2026 | vLLM is an inference and serving engine for large language models. Prior to 0.30.0, OpenAI-compatible request models accept a non-empty cache_salt value without enforcing the character and length restrictions required by the IPCCacheServerKey consumer in LMCache-MP. On… | ||
| CVE-2026-105754 | Med | 0.35 | 6.5 | 0.00 | Oct 5, 2026 | vLLM is an inference and serving engine for large language models. Prior to 0.30.0, the /inference/v1/generate endpoint in the disaggregated scale-out path accepts caller-supplied tensors in the features.kwargs_data field, cache identifiers in the features.mm_hashes field,… | ||
| CVE-2026-105753 | Med | 0.35 | 6.5 | 0.00 | Oct 5, 2026 | vLLM is an inference and serving engine for large language models. Prior to 0.28.0, the default mirrored multimodal LRU cache can commit a media hash in the frontend sender cache during multimodal rendering and before engine admission, while the engine receiver cache never… | ||
| CVE-2026-100654 | Med | 0.35 | 6.5 | 0.00 | Sep 26, 2026 | vLLM before 0.29.0 accepts user-controlled stop_token_ids on the OpenAI-compatible POST /v1/completions and POST /v1/chat/completions endpoints but validates only that the values are integers, not that each token id is within the model vocabulary/logits range. When min_tokens >… | ||
| CVE-2026-100653 | Med | 0.35 | 6.5 | 0.00 | Sep 26, 2026 | 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 several Hugging Face artifact loads for the FunAudioChat and Tarsier2… | ||
| CVE-2026-100651 | Med | 0.35 | 6.5 | 0.00 | Sep 26, 2026 | vLLM before 0.29.0 fails to enforce decoder prompt-length validation on the disaggregated serving endpoint /inference/v1/generate. When the request contains a 'features' (multimodal) payload, vllm/entrypoints/serve/disagg/serving.py builds a multimodal EngineInput directly from… | ||
| CVE-2026-100650 | Med | 0.35 | 6.5 | 0.01 | Sep 26, 2026 | 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 --limit-mm-per-prompt item limits). Across four ingress… | ||
| CVE-2026-69147 | Med | 0.35 | 6.5 | 0.01 | Sep 16, 2026 | vLLM is an inference and serving engine for large language models. Prior to 0.28.0, request bodies for Chat Completions and Responses can set media_io_kwargs.video.video_backend to pynvvideocodec, and MediaConnector.fetch_video forwards that choice to VideoMediaIO even when… | ||
| CVE-2026-57173 | Med | 0.35 | 6.5 | 0.01 | Sep 16, 2026 | vLLM is an inference and serving engine for large language models. Prior to 0.24.0, the input_audio handling path for /v1/chat/completions calls AudioMediaIO.load_bytes or AudioMediaIO.load_file without passing VLLM_MAX_AUDIO_DECODE_DURATION_S to the shared audio decoder. An… | ||
| CVE-2026-90555 | Med | 0.35 | 6.5 | 0.01 | Sep 12, 2026 | vLLM versions before 0.28.0 fail to validate audio sample rate headers in the transcription endpoint, allowing authenticated clients to bypass duration checks. Attackers can submit forged FLAC headers with inflated sample rates to trigger excessive memory allocation and crash… | ||
| CVE-2026-73560 | Med | 0.35 | 6.5 | 0.00 | Aug 17, 2026 | vLLM is an inference and serving engine for large language models. Prior to 0.26.0, the MiMoV2OmniMultiModalProcessor in vllm/transformers_utils/processors/mimo_v2_omni.py passes attacker-controlled image and audio strings through _fetch_image, requests.get, and Image.open… | ||
| CVE-2026-73559 | Med | 0.35 | 6.5 | 0.01 | Aug 13, 2026 | vLLM is an inference and serving engine for large language models. From 0.19.0 until 0.26.0, the /v1/completions CompletionRequest.prompt field in vllm/entrypoints/openai/completion/protocol.py accepts an unbounded list[str] or list[list[int]], prompt_to_seq() in… | ||
| CVE-2026-55514 | Med | 0.35 | 6.5 | 0.01 | Jul 6, 2026 | vLLM is a library for LLM inference and serving. From 0.12.0 to before 0.24.0, sending a pure prompt embeds payload in a /v1/completions request with a model using M-RoPE causes EngineCore to fail an assertion and fatally crash, shutting down the entire server application. Any… | ||
| CVE-2026-55646 | Med | 0.35 | 6.5 | 0.01 | Jul 6, 2026 | vLLM is an inference and serving engine for large language models. From 0.22.0 to 0.23.0, the /v1/audio/transcriptions and /v1/audio/translations routes call request.file.read() to fully materialize an uploaded audio file into memory before vLLM checks the documented… |
- risk 0.35cvss 6.5epss 0.00
vLLM is an inference and serving engine for large language models. From 0.19.0 until 0.32.0, authenticated callers can supply a request chat template when no server-side --chat-template is configured, or a malicious model repository can provide a template that is compiled and…
- risk 0.35cvss 6.5epss 0.00
vLLM from 0.20.1 before 0.29.0 contains an improper input validation vulnerability that allows remote attackers to crash the engine by supplying out-of-vocabulary stop_token_ids. Attackers can send OpenAI chat, completion, generate, or batch requests with invalid IDs to…
- risk 0.35cvss 6.5epss 0.00
vLLM from 0.19.0 before 0.24.0 contains a decompression bomb vulnerability because the AudioMediaIO multimodal path calls load_audio without the max_duration_s decoded-duration limit. Authenticated attackers can send small compressed audio via input_audio or audio_url in…
- risk 0.35cvss 6.5epss 0.00
vLLM before 0.31.0 contains a cache aliasing vulnerability that allows authenticated attackers to collide shared cache keys by supplying caller-controlled values such as EXIF ImageID or cache_salt. Attackers can set a UUID EXIF ImageID to receive another request's cached…
- risk 0.35cvss 6.5epss 0.00
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. Authenticated remote attackers can repeatedly upload malformed IAMF containers to /v1/audio/transcriptions, corrupting the…
- risk 0.35cvss 6.5epss 0.00
vLLM is an inference and serving engine for large language models. Prior to 0.30.0, structured-output request failures can escape request-scoped validation and reach the EngineCore fatal-error path. A per-request backend mismatch can re-raise a grammar compilation exception,…
- risk 0.35cvss 6.5epss 0.00
vLLM is an inference and serving engine for large language models. Prior to 0.30.0, OpenAI-compatible request models accept a non-empty cache_salt value without enforcing the character and length restrictions required by the IPCCacheServerKey consumer in LMCache-MP. On…
- risk 0.35cvss 6.5epss 0.00
vLLM is an inference and serving engine for large language models. Prior to 0.30.0, the /inference/v1/generate endpoint in the disaggregated scale-out path accepts caller-supplied tensors in the features.kwargs_data field, cache identifiers in the features.mm_hashes field,…
- risk 0.35cvss 6.5epss 0.00
vLLM is an inference and serving engine for large language models. Prior to 0.28.0, the default mirrored multimodal LRU cache can commit a media hash in the frontend sender cache during multimodal rendering and before engine admission, while the engine receiver cache never…
- risk 0.35cvss 6.5epss 0.00
vLLM before 0.29.0 accepts user-controlled stop_token_ids on the OpenAI-compatible POST /v1/completions and POST /v1/chat/completions endpoints but validates only that the values are integers, not that each token id is within the model vocabulary/logits range. When min_tokens >…
- risk 0.35cvss 6.5epss 0.00
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 several Hugging Face artifact loads for the FunAudioChat and Tarsier2…
- risk 0.35cvss 6.5epss 0.00
vLLM before 0.29.0 fails to enforce decoder prompt-length validation on the disaggregated serving endpoint /inference/v1/generate. When the request contains a 'features' (multimodal) payload, vllm/entrypoints/serve/disagg/serving.py builds a multimodal EngineInput directly from…
- risk 0.35cvss 6.5epss 0.01
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 --limit-mm-per-prompt item limits). Across four ingress…
- risk 0.35cvss 6.5epss 0.01
vLLM is an inference and serving engine for large language models. Prior to 0.28.0, request bodies for Chat Completions and Responses can set media_io_kwargs.video.video_backend to pynvvideocodec, and MediaConnector.fetch_video forwards that choice to VideoMediaIO even when…
- risk 0.35cvss 6.5epss 0.01
vLLM is an inference and serving engine for large language models. Prior to 0.24.0, the input_audio handling path for /v1/chat/completions calls AudioMediaIO.load_bytes or AudioMediaIO.load_file without passing VLLM_MAX_AUDIO_DECODE_DURATION_S to the shared audio decoder. An…
- risk 0.35cvss 6.5epss 0.01
vLLM versions before 0.28.0 fail to validate audio sample rate headers in the transcription endpoint, allowing authenticated clients to bypass duration checks. Attackers can submit forged FLAC headers with inflated sample rates to trigger excessive memory allocation and crash…
- risk 0.35cvss 6.5epss 0.00
vLLM is an inference and serving engine for large language models. Prior to 0.26.0, the MiMoV2OmniMultiModalProcessor in vllm/transformers_utils/processors/mimo_v2_omni.py passes attacker-controlled image and audio strings through _fetch_image, requests.get, and Image.open…
- risk 0.35cvss 6.5epss 0.01
vLLM is an inference and serving engine for large language models. From 0.19.0 until 0.26.0, the /v1/completions CompletionRequest.prompt field in vllm/entrypoints/openai/completion/protocol.py accepts an unbounded list[str] or list[list[int]], prompt_to_seq() in…
- risk 0.35cvss 6.5epss 0.01
vLLM is a library for LLM inference and serving. From 0.12.0 to before 0.24.0, sending a pure prompt embeds payload in a /v1/completions request with a model using M-RoPE causes EngineCore to fail an assertion and fatally crash, shutting down the entire server application. Any…
- risk 0.35cvss 6.5epss 0.01
vLLM is an inference and serving engine for large language models. From 0.22.0 to 0.23.0, the /v1/audio/transcriptions and /v1/audio/translations routes call request.file.read() to fully materialize an uploaded audio file into memory before vLLM checks the documented…
Page 4 of 8