CVE-2026-105755
Fixed in a distro release EPSS <1%CVSS 4.2
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.
Products: vllm-project vllm
Fixed upstream
| Ecosystem | Package | Fixed in |
|---|---|---|
| PyPI | vllm | 0.30.0 |
Distro releases
| Release | Package | Fixed in |
|---|---|---|
| Chainguard | kserve-huggingfaceserver-cuda-13.3 | no fix |
| Chainguard | lmcache-cuda-12.8 | no fix |
| Chainguard | nemo-rl-cuda-13.0-vllm | no fix |
| Chainguard | py3.10-vllm-cuda-12.4 | no fix |
| Chainguard | py3.12-vllm-cuda-12.4 | no fix |
| Chainguard | tritonserver-backend-vllm-cuda-13.0 | 25.11-r17 |
| Chainguard | vllm-cuda-13.2 | 0.32.0-r0 |
| Chainguard | vllm-openai-cuda-13.0 | 0.31.0-r0 |