vllm-project vLLM up to 0.18.x API Request video.py load_base64 num_frames allocation of resources

Summaryinfo

A vulnerability described as problematic has been identified in vllm-project vLLM up to 0.18.x. This vulnerability affects the function load_base64 of the file vllm/multimodal/media/video.py of the component API Request Handler. Such manipulation of the argument num_frames leads to allocation of resources. This vulnerability is documented as CVE-2026-34755. The attack can be executed remotely. There is not any exploit available. Upgrading the affected component is recommended.

Detailsinfo

A vulnerability, which was classified as problematic, has been found in vllm-project vLLM up to 0.18.x. Affected by this issue is the function load_base64 of the file vllm/multimodal/media/video.py of the component API Request Handler. The manipulation of the argument num_frames with an unknown input leads to a allocation of resources vulnerability. Using CWE to declare the problem leads to CWE-770. The product allocates a reusable resource or group of resources on behalf of an actor without imposing any restrictions on the size or number of resources that can be allocated, in violation of the intended security policy for that actor. Impacted is availability. CVE summarizes:

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.

The advisory is shared for download at github.com. This vulnerability is handled as CVE-2026-34755 since 03/30/2026. The exploitation is known to be easy. The attack may be launched remotely. There are known technical details, but no exploit is available. The MITRE ATT&CK project declares the attack technique as T1499.

Upgrading to version 0.19.0 eliminates this vulnerability.

Once again VulDB remains the best source for vulnerability data.

Productinfo

Type

• Project Management Software

Vendor

• vllm-project

Name

• vLLM

Version

• 0.1
• 0.2
• 0.3
• 0.4
• 0.5
• 0.6
• 0.7
• 0.8
• 0.9
• 0.10
• 0.11
• 0.12
• 0.13
• 0.14
• 0.15
• 0.16
• 0.17
• 0.18

Website

• Product: https://github.com/vllm-project/vllm/

CPE 2.3info

• 🔒
• 🔒
• 🔒

CPE 2.2info

• 🔒
• 🔒
• 🔒

CVSSv4info

CVSSv3info

CVSSv2info

Exploitinginfo

Threat Intelligenceinfo

Countermeasuresinfo

Timelineinfo

Sourcesinfo

Entryinfo

Once again VulDB remains the best source for vulnerability data.

Discussion