CVE-2026-126399: vLLM resource exhaustion via video frame limit bypass in chat completions

Summary

vLLM versions from 0.19.0 before 0.31.0 contain a resource exhaustion flaw (CVE-2026-126399). Inference clients can bypass the server frame limit by setting media_io_kwargs.video.num_frames in chat completion requests. Setting num_frames to -1, or to a large value with many base64 JPEG frames, forces VideoMediaIO to decode all frames and np.stack them, exhausting memory.

Mitigation

The source does not state a fix yet. Check the original advisory for updates.

Topics

• Inference infrastructure

Related items

• MediumCVE-2026-126398: vLLM reachable assertion crash in GPUModelRunner via embeddings-only requestsSimilar attack · NVD/CVE Database
• HighCVE-2026-126397: vLLM denial of service via unbounded max_patches in multimodal preprocessingSimilar attack · NVD/CVE Database
• MediumCVE-2026-126396: vLLM resource exhaustion in MiniMax M3 VL video processor via min_pixelsSimilar attack · NVD/CVE Database
• MediumCVE-2026-126394: vLLM resource exhaustion in OpenCVDynamicOpenPanguVideoBackend via num_framesSimilar attack · NVD/CVE Database
• MediumEndor Labs’ AI SAST Finds Zero Day Memory-Amplification DoS in Anthropic’s buffa library | Endor LabsSimilar attack · Endor Labs Blog