6.5Medium
CVE-2026-116181vLLM 0.20.1 before 0.29.0 DoS and GPU Out-of-Bounds Write via stop_token_ids
Description
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.
Weakness (CWE):
CWE-129
Data: CVE.org, NVD, CISA KEV, FIRST EPSS, Ubuntu. Published as the sources state it, in English.
Affected products and versions
| Vendor | Product | Versions | |
|---|---|---|---|
| vllm-project | vllm | 0.20.1 – < 0.29.0, fixed 0.29.0 |
References and patches
- GitHub Security Advisory (GHSA-w9r6-29q4-pvpr) vendor-advisoryPatchVendor Advisory
- VulnCheck Advisory: vLLM 0.20.1 before 0.29.0 DoS and GPU Out-of-Bounds Write via stop_token_ids third-party-advisoryThird Party Advisory
CPE matches (NVD) (1)
cpe:2.3:a:vllm:vllm:*:*:*:*:*:*:*:*≥ 0.20.1 < 0.29.0