CVE-2026-109228 sgl-project CVE debrief

Who should care

Defenders responsible for SGLang deployments, especially those processing video content, should assess exposure and prioritize patching or mitigation. This includes operators managing SGLang instances, platform administrators, vulnerability management teams, and security teams responsible for monitoring and incident response.

Why it matters

CVE-2026-109228 is a high-severity vulnerability in SGLang that allows remote attackers to bypass resource budget limits via GLM-V video parameters, potentially leading to a denial-of-service (DoS) attack. Defenders should prioritize patching or mitigating this vulnerability, especially in environments where SGLang is used to process video content.

• Potential denial-of-service (DoS) attacks by exhausting worker memory
• Bypass of resource budget limits via GLM-V video parameters
• Possible impact on video processing services
• Need for validation of user-supplied video parameters

Technical summary

The vulnerability is caused by a lack of proper validation of user-supplied video parameters in SGLang, which allows attackers to override GLM-V video processing limits. This can lead to a DoS attack by exhausting worker memory. The vulnerability has a CVSS score of 8.7 and can be exploited by supplying specially crafted fps, max_frames, max_tokens_per_frame, and max_image_tokens values in the video_url object, potentially leading to a denial-of-service (DoS) attack by exhausting worker memory. Defenders should prioritize patching or mitigating this vulnerability, especially in environments where SGLang is used to process video content.

Defensive priority

Defenders should prioritize patching or mitigating this vulnerability, especially in environments where SGLang is used to process video content.

Recommended defensive actions

• Patch SGLang to version greater than 0.5.21
• Restrict access to SGLang to trusted users only
• Monitor SGLang logs for suspicious activity
• Implement rate limiting on video processing requests
• Review compensating controls for exposed systems while remediation is scheduled and verified
• Check relevant monitoring, detection, and logs for exposed assets that need extra review
• Track exceptions, retest remediated assets, and close the item only after evidence is documented.

Evidence notes

The vulnerability is caused by a lack of proper validation of user-supplied video parameters, which allows attackers to override GLM-V video processing limits. This can lead to a DoS attack by exhausting worker memory.

Sources and references

Methodology and review provenance

AI-assisted synthesis based on stored public vulnerability evidence. System validation, approval state, and publication status do not by themselves establish human review of this revision. PatchSiren helps prioritize defensive review and does not prove exposure or remediation on any system.