CVE-2026-34755 PUBLISHED

vLLM Affected by Denial of Service via Unbounded Frame Count in video/jpeg Base64 Processing

Assigner: GitHub_M
Reserved: 30.03.2026 Published: 06.04.2026 Updated: 06.04.2026

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.

Metrics

CVSS Vector: CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:H
CVSS Score: 6.5

Product Status

Vendor vllm-project
Product vllm
Versions
  • Version >= 0.7.0, < 0.19.0 is affected

References

Problem Types

  • CWE-770: Allocation of Resources Without Limits or Throttling CWE