CVE-2026-78684 vllm-project CVE debrief

Who should care

Defenders responsible for vLLM deployments, particularly those using the DeepStream GPU backend, should assess exposure and prioritize verification and remediation efforts. Defenders should verify vLLM versions, assess exposure to DeepStream GPU backend usage, and implement resource controls and monitoring. The vulnerability affects vLLM before 0.27.0 and allows unauthenticated attackers to cause partial denial of service.

Why it matters

CVE-2026-78684 is a medium-severity vulnerability in vLLM that allows unauthenticated attackers to cause partial denial of service. Defenders should prioritize verifying vLLM versions, assessing exposure to DeepStream GPU backend usage, and implementing resource controls and monitoring.

• Partial denial of service for concurrent requests
• Potential for resource exhaustion if not properly mitigated
• Need for verification of vLLM version and exposure to DeepStream GPU backend usage
• Importance of implementing resource controls and monitoring

Technical summary

vLLM before 0.27.0 fails to properly classify DeepStream as a GPU backend, allowing unauthenticated attackers to bypass resource controls and cause partial denial of service. The vulnerability can be triggered by activating DeepStream at request time to initialize the process-wide GPU decode pool and submit video that bypasses resource controls. The CVE record and NVD entry provide details on the vulnerability, but additional information on affected versions and remediation is needed. Defenders should prioritize verifying vLLM versions and assessing exposure to DeepStream GPU backend usage.

Defensive priority

Defenders should prioritize verifying vLLM versions and assessing exposure to DeepStream GPU backend usage.

Recommended defensive actions

• Verify vLLM version and assess exposure to DeepStream GPU backend usage
• Implement resource controls and monitoring for concurrent requests
• Review and update incident response plans for potential denial of service scenarios
• 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
• Confirm whether affected product deployments exist in managed environments and assign an owner for follow-up
• Review compensating controls for exposed systems while remediation is scheduled and verified

Evidence notes

The CVE record and NVD entry provide details on the vulnerability, but additional information on affected versions and remediation is needed. The vulnerability affects vLLM before 0.27.0 and allows unauthenticated attackers to cause partial denial of service. Defenders should verify vLLM versions, assess exposure to DeepStream GPU backend usage, and implement resource controls and monitoring. The source detail is limited, and explicit evidence-limit language and defensive verification tasks are necessary.

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.