CVE-2026-34755

vLLM is an inference and serving engine for large language models (LLMs). From 0.7.0 to before 0.19.0, the VideoMediaIO.load_base64() method at vllm/multimodal/media/video.py splits video/jpeg data URLs by comma to extract individual JPEG frames, but does not enforce a frame count limit. The num_frames parameter (default: 32), which is enforced by the load_bytes() code path, is completely bypassed in the video/jpeg base64 path. An attacker can send a single API request containing thousands of comma-separated base64-encoded JPEG frames, causing the server to decode all frames into memory and crash with OOM. This vulnerability is fixed in 0.19.0.
Configurations

Configuration 1 (hide)

cpe:2.3:a:vllm:vllm:*:*:*:*:*:*:*:*

History

No history.

Information

Published : 2026-04-06 16:16

Updated : 2026-08-25 13:18


NVD link : CVE-2026-34755

Mitre link : CVE-2026-34755

CVE.ORG link : CVE-2026-34755


JSON object : View

Products Affected

vllm

  • vllm
CWE
CWE-770

Allocation of Resources Without Limits or Throttling