HIGHCVSS v3.1 7.7Vulnerability
CVE-2026-116180
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.
Properties
- Severity
- HIGH
- Nvd fetched at
- 2026-10-25T14:48:34.194958+00:00
- Cvss severity
- HIGH
- Cvss score
- 7.7
- Retrieved at
- 2026-10-25T14:49:38+00:00
- Score
- 7.7
- Cvss version
- 3.1
- Last source
- NVD
- Cve id
- CVE-2026-116180
- Cvss vector
- CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:C/C:N/I:N/A:H
- First seen by Signal
- 2026-10-25T14:49:38+00:00
- Vector
- CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:C/C:N/I:N/A:H
- Published at
- 2026-10-25T14:16:57.760
- Last modified
- 2026-10-25T14:16:57.760
Related Entities (1)
HAS_WEAKNESS (1)
Explore deeper with Ninja Signal's threat intelligence graph