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

Vulnerability summary

CVE-2026-90554 is a high (CVSS 7.5) Uncontrolled Resource Consumption vulnerability in vllm-project vllm. EPSS estimates a 0.20% chance of exploitation in the next 30 days, higher than 9.4% of scored vulnerabilities.

Timeline

1. 12-09-2026
   PublishedVulnerability first disclosed.
2. 14-09-2026
   Last modifiedVulnerability information updated.

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.

• CWE-400Uncontrolled Resource ConsumptionThe product does not properly control the allocation and maintenance of a limited resource.