CVE-2026-5497
HighCVSS 7.5Exploitation Probability (EPSS)
Low risk42th percentile - higher than 42% of all known CVEs
Summary
A vulnerability in vLLM versions 0.8.0 and later allows an OOM DoS attack via the `VideoMediaIO.load_base64()` method. An attacker can send a single API request with thousands of base64-encoded JPEG frames, causing server crash due to memory exhaustion.
Risk Assessment
The organization faces service disruption from an unauthenticated remote attacker who can crash the vLLM server using the publicly accessible OpenAI-compatible API.
Recommendation
Immediately update vLLM to a patched version that limits video frame count, or deploy WAF rules to block requests with excessive JPEG frames in data URLs.
Original NVD description (English source)
vLLM versions 0.8.0 and later are vulnerable to an Out-of-Memory (OOM) Denial of Service (DoS) attack due to unbounded frame count processing in the `VideoMediaIO.load_base64()` method. When processing `video/jpeg` data URLs, the method splits the base64 data string on commas to extract individual JPEG frames without enforcing a frame count limit. An attacker can exploit this by crafting a single API request containing thousands of comma-separated base64-encoded JPEG frames in a data URL, causing the server to decode all frames into memory and crash due to excessive memory consumption. This vulnerability is reachable via the OpenAI-compatible chat completions API and does not require authentication.