Vulnerability · Published Sep 12, 2026 · Assigned by VulnCheck

CVE-2026-90554vLLM before 0.28.0 Denial of Service via audio extraction

Medium 6.2

In plain English

A medium-severity uncontrolled resource consumption in vLLM from vllm-project. There are no confirmed reports of exploitation so far. The chance it is exploited in the next 30 days is estimated at 0.2% (higher than 10% of all known vulnerabilities).

Severity (CVSS 3.1)6.2Medium

Exploit odds (30 days)0.2%top 90.0% of all CVEs

Exploited in the wildNo reportsCISA SSVC: none

Attack toolingNoneno public tooling known

Official description

vLLM versions >=0.10.2 and <0.28.0 do not apply any audio decode-size or duration limit when extracting audio from video input for NanoNemotronVL models. In nano_nemotron_vl.py, _extract_audio_from_videos calls load_audio_pyav(BytesIO(video_bytes)) without the max_duration_s or max_decode_bytes parameters, so neither VLLM_MAX_AUDIO_DECODE_DURATION_S nor VLLM_MAX_AUDIO_DECODE_BYTES is enforced (unlike the direct audio upload path in AudioMediaIO). When a NanoNemotronVL model is served with use_audio_in_video=True, an attacker who supplies a small, highly compressed video as multimodal input can force the server to allocate gigabytes of memory during audio decoding, resulting in a denial of service. Fixed in vLLM 0.28.0.

Affected products (as reported)

Vendor	Product	Versions
vllm-project	vLLM	0.10.2 – before 0.28.0

Weakness: CWE-400 · Vector: CVSS:3.1/AV:L/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H

References

GitHub Security Advisory (GHSA-936p-m5pv-vvjf) vendor-advisory
VulnCheck Advisory: vLLM before 0.28.0 Denial of Service via audio extraction third-party-advisory