Improper Handling of Highly Compressed Data (Data Amplification) Affecting vllm package, versions [0.19.0,0.24.0)
Threat Intelligence
The probability is the direct output of the EPSS model, and conveys an overall sense of the threat of exploitation in the wild. The percentile measures the EPSS probability relative to all known EPSS scores. Note: This data is updated daily, relying on the latest available EPSS model version. Check out the EPSS documentation for more details.
Do your applications use this vulnerable package?
In a few clicks we can analyze your entire application and see what components are vulnerable in your application, and suggest you quick fixes.
Test your applications- Snyk IDSNYK-PYTHON-VLLM-21707827
- published25 Oct 2026
- disclosed25 Oct 2026
- creditRex Liu
Introduced: 25 Oct 2026
NewCVE-2026-116175 (opens in a new tab)How to fix?
Upgrade vllm to version 0.24.0 or higher.
Overview
vllm is an A high-throughput and memory-efficient inference and serving engine for LLMs
Affected versions of this package are vulnerable to Improper Handling of Highly Compressed Data (Data Amplification) via the AudioMediaIO multimodal path, which calls load_audio without passing a max_duration_s decoded-duration limit. An authenticated attacker can submit a small compressed audio payload via input_audio data URLs or audio_url content parts in /v1/chat/completions requests, causing the decoder to accumulate decoded PCM frames until server memory is exhausted. The --allowed-media-domains option, media fetch timeouts, and HTTP body limits do not enforce a decoded PCM duration budget, so the issue remains reachable for allowed audio_url domains and for input_audio data that fits within normal request size limits. This is a sibling incomplete-fix variant of CVE-2026-54233, which added the duration guard only to the speech-to-text path.