Security Vulnerabilities
vLLM 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 a decode address in the X-Request-Id header. Attackers can make a P2pNcclConnector prefill worker ZMQ-connect to an attacker host, leaking the NCCL unique_id and KV-cache tensors or blocking the worker.
CVSS Score
8.8
EPSS Score
0.003
Published
2026-10-25
vLLM from 0.10.2 before 0.25.0 contains an uncontrolled resource consumption vulnerability that allows unauthenticated attackers to exhaust memory because the Nemotron-VL processors globally set Image.MAX_IMAGE_PIXELS to None. Attackers can submit a pathological decompression-bomb image to /v1/chat/completions to OOM-kill the worker and deny service to co-tenants.
CVSS Score
8.7
EPSS Score
0.003
Published
2026-10-25
vLLM from 0.20.0 before 0.30.0 contains a denial of service vulnerability that allows remote attackers to crash the engine by sending inconsistent multimodal metadata to the disaggregated /inference/v1/generate endpoint. Attackers can replay a valid multimodal item with repeated or overlapping mm_placeholders ranges or mismatched kwargs_data and mm_hashes lists, producing a negative text span in Qwen2.5-VL position construction that terminates the engine.
CVSS Score
8.2
EPSS Score
0.003
Published
2026-10-25
vLLM from 0.18.0 before 0.31.0 contains a resource exhaustion vulnerability that allows API clients to bypass the video/jpeg frame-count cap by overriding num_frames through media_io_kwargs. Attackers can set num_frames to -1 with a large inline video so VideoMediaIO.load_base64 decodes every frame, exhausting memory and crashing the API frontend.
CVSS Score
8.7
EPSS Score
0.003
Published
2026-10-25
vLLM from 0.8.5 before 0.31.0 has a denial of service flaw. LogprobsTensors.empty_cpu() in vllm/v1/outputs.py allocates an uninitialized int32 tensor, and the V1 model runner leaves prompt positions reported as computed by a KV connector unwritten. A client sharing the engine can send prompt_logprobs (or echo with logprobs) against a connector-cached prefix, making tokenizer decode raise OverflowError, which crashes EngineCore and drops all co-tenant requests.
CVSS Score
8.3
EPSS Score
0.003
Published
2026-10-25
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 terminate the engine or write outside the V2 runner's logits buffer.
CVSS Score
7.1
EPSS Score
0.004
Published
2026-10-25
vLLM from 0.22.0 before 0.24.0 contains an improper input validation vulnerability in the Rust gRPC converter that allows remote attackers to bypass max_logprobs limits via CandidateTokens.top_n. Attackers can send Generate or GenerateStream requests with oversized or integer-wrapping top_n values to trigger torch.topk failures that terminate EngineCore.
CVSS Score
8.2
EPSS Score
0.003
Published
2026-10-25
vLLM from 0.22.0 before 0.24.0 contains a denial of service vulnerability that allows remote attackers to terminate EngineCore by bypassing logprob_token_ids limits in the Rust gRPC frontend. Attackers can send Generate or GenerateStream requests with output_logprobs enabled and more than 128 output_candidates.token_ids, triggering a fatal sampler exception that crashes the engine.
CVSS Score
8.2
EPSS Score
0.003
Published
2026-10-25
vLLM from 0.22.0 before 0.24.0 contains an improper input validation vulnerability in the Rust gRPC Generate and GenerateStream service that forwards unchecked prompt token IDs to EngineCore. Attackers can submit requests with out-of-vocabulary prompt.token_ids values to trigger a fatal embedding lookup failure that terminates EngineCore and leaves the server unhealthy.
CVSS Score
8.2
EPSS Score
0.003
Published
2026-10-25
A cross-site request forgery vulnerability in the OAuth extension of Zimbra Collaboration Suite (ZCS) Network Edition allows an attacker to link an attacker-controlled external account to a victim's mailbox. The OAuth callback does not bind its state value or authorization code to the initiating Zimbra session and instead selects the destination mailbox from the browser's current authentication cookie. On a deployment with the provider configured, an attacker can induce a signed-in victim to load a callback containing a fresh code for the attacker's provider account. This creates a persistent data source that imports attacker-controlled contacts or events. With bidirectional CalDAV synchronization, appointments subsequently created or edited in the linked calendar are also disclosed to the attacker. ZCS 10.1.16 and 10.1.20 are affected.
CVSS Score
7.1
EPSS Score
0.002
Published
2026-10-23