CVE intelligence and bounded remediation
CVE-2025-46560 — vLLM is a high-throughput and memory-efficient inference and serving engine for LLMs
vLLM is a high-throughput and memory-efficient inference and serving engine for LLMs. Versions starting from 0.8.0 and prior to 0.8.5 are affected by a critical performance vulnerability in the input preprocessing logic of the multimodal tokenizer. The code dynamically replaces placeholder tokens (e.g., <|audio_|>, <|image_|>) with repeated tokens based on precomputed lengths. Due to inefficient list concatenation operations, the algorithm exhibits quadratic time complexity (O(n²)), allowing malicious actors to trigger resource exhaustion via specially crafted inputs. This issue has been patched in version 0.8.5.
- Severity
- High
- CVSS
- 7.5 (3.1)
- Published
- 2025-04-30
- CISA KEV
- Not currently listed
- Ecosystem
- software/application
- Weaknesses
- CWE-1333
Affected products
- vllm / vllm
Matched remediation archetype
General vulnerability remediation
This catalog composition supplies bounded fallback guidance. Explicitly reviewed curated workflows load with the complete record below.
Check exposure
- Confirm the affected component, deployment paths, reachable interfaces, and enabled features from inventories and configuration, without probing production destructively.
- Compare the advisory's affected conditions with the repository lockfiles, build manifests, artifacts, and runtime inventory.
- Identify data sensitivity, trust boundaries, and privilege level for every confirmed affected deployment.
Remediate safely
- Apply a vendor-supported fix or remove the affected component or feature; record the selected change and its source in the repository.
- Update direct and transitive dependency locks, generated artifacts, deployment manifests, and asset inventories together.
- Add a regression test for the documented unsafe condition using inert inputs and preserve rollback instructions.
Authoritative sources
Complete CVE record and remediation plan
The detailed catalog view below loads this exact record, its source evidence, and the full seven-phase agentic change plan.