Description
vLLM is an inference and serving engine for large language models. Prior to 0.30.0, flash late-interaction scoring at the /score and /rerank endpoints derives each worker's query_key value from the caller-controlled X-Request-Id header. A concurrent request that reuses a victim's identifier can overwrite the cached query embedding so the victim's documents are scored against the attacker's query, and shared use counters can also cause a late-interaction cache-miss error. This issue is fixed in version 0.30.0.
Metrics
CVSS:3.1/AV:N/AC:H/PR:L/UI:N/S:U/C:N/I:L/A:LWeakness enumeration
Affected software
| Vendor | Product | Versions |
|---|---|---|
| vllm-project | vllm | < 030.0 |
References
- https://github.com/vllm-project/vllm/commit/ee17d0d869203ef9a35ad73358a4987bba14b1fcPatch
- https://github.com/vllm-project/vllm/pull/51445Issue Tracking, Patch
- https://github.com/vllm-project/vllm/releases/tag/v0.30.0Release Notes
- https://github.com/vllm-project/vllm/security/advisories/GHSA-2phq-3phc-84pxVendor Advisory
Timeline
- Published
- Oct 5, 2026
- Last modified
- Oct 8, 2026
- NVD status
- Analyzed