CVE-2026-100648: vllm before 0.29.0 Uncontrolled Resource Consumption via Audio Decoding
vllm before 0.29.0 fails to enforce VLLMMAXAUDIOCLIPFILESIZEMB limit in multimodal chat audio decoding, allowing unauthenticated clients to bypass file size restrictions. Attackers can submit oversized audio files through chat endpoints to consume excessive memory and CPU resources during decoding.
Affected Software
Remediation
Recommended actions to resolve this vulnerability, in priority order.
- Upgrade
Upgrade
vllmto a version that resolves this vulnerability.Fixed in 0.29.0
Event History
Frequently Asked Questions
Who is exposed to this issue?
Deployments of vllm before 0.29.0 that expose multimodal chat endpoints capable of accepting audio input are affected. Unauthenticated clients can submit the malicious input if they can reach those endpoints.
What does an attacker need to exploit it?
An attacker only needs network access to a vulnerable chat endpoint and the ability to submit an oversized audio file. No authentication or user interaction is required.
Does setting VLLM_MAX_AUDIO_CLIP_FILESIZE_MB protect vulnerable versions?
No. Versions before 0.29.0 fail to enforce the VLLM_MAX_AUDIO_CLIP_FILESIZE_MB limit during multimodal chat audio decoding, so oversized files can bypass that restriction.
What is the impact of successful exploitation?
Oversized audio submissions can consume excessive memory and CPU during decoding, potentially reducing availability of the affected service.