Vllm-project vLLM
vLLM before 0.28.0 Denial of Service via audio extraction
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 versions and mitigation
From the CVE record and the vendor's references.Affected
- >= 0.10.2, < 0.28.0
Fixed in 0.28.0
Mitigation
Upgrade vLLM to 0.28.0 or later.
References
- Vendor advisory github.com/vllm-project/vllm/security/advisori…
Timeline
Days relative to CVE publicationNo public PoC found yet; not in CISA KEV.
PoC repositories
Found on GitHub.No PoC repositories found yet.