CVE-2026-105922 vllm-project CVE debrief

Who should care

Defenders of systems using vLLM up to 0.31.0 should assess exposure and prioritize patching or mitigation. This includes operators, platform administrators, vulnerability management teams, and security teams who need to verify exposure, assess potential impact, and apply patches or mitigations as available. They should also monitor for public exploit usage and review compensating controls for exposed systems.

Why it matters

CVE-2026-105922 is a denial-of-service vulnerability in vLLM up to 0.31.0. Defenders should verify exposure, assess potential impact, and prioritize patching or mitigation.

• Potential denial-of-service impact on vLLM installations
• Need for verification of exposure and potential exploitation
• Priority for applying patches or mitigations
• Monitoring for public exploit usage

Technical summary

The vulnerability affects vLLM up to 0.31.0, specifically the get_token_bin_counts_and_mask function in vllm/model_executor/layers/utils.py. Remote exploitation is possible, leading to denial-of-service. Defenders should assess the potential impact on vLLM installations and prioritize patching or mitigation. The exploit has been released to the public and may be used for attacks. The project was informed of the problem early through an issue report but has not responded yet. Affected product deployments should be identified in managed environments, and owners should be assigned for follow-up.

Defensive priority

Defenders should prioritize verifying exposure of vLLM installations, assessing the impact of potential denial-of-service attacks, and applying patches or mitigations as available.

Recommended defensive actions

• Verify vLLM installations for exposure
• Assess potential denial-of-service impact
• Apply patches or mitigations as available
• Monitor for public exploit usage
• 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 source corpus provides details on the vulnerability in vLLM up to 0.31.0, including the affected function and potential denial-of-service impact. However, some details like specific exploitation scenarios or victim impact are not provided.

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.