CVE-2026-52132 ggml CVE debrief

Who should care

Organizations using ggml's Llama.cpp, especially those exposing the /rerank endpoint to untrusted sources, should prioritize patching to prevent potential denial-of-service attacks. This includes reviewing system configurations, ensuring the --reranking flag is not used unnecessarily, and implementing compensating controls for exposed systems while remediation is scheduled and verified. Additionally, security teams should monitor for unusual traffic patterns or errors related to the --reranking flag and review relevant logs for exposed assets that need extra review. Operators and platform administrators should also be aware of the potential impact on their systems and plan accordingly for updates or mitigations through normal change control where exposure is confirmed. Vulnerability management teams should assess the affected scope and severity based on the official advisory or CVE record to validate the necessary actions for their environments. This involves confirming whether affected product deployments exist in managed environments and assigning an owner for follow-up actions to ensure timely remediation and minimize potential operational impact. The affected product or component is ggml's Llama.cpp, and the vulnerability class is a denial-of-service attack. The likely operational impact includes service disruption due to std::bad_alloc and HTTP 500 errors. Source-confidence limits are based on the CVE record and related advisories, and the review context should include system configurations, traffic patterns, and error logs related to the --reranking flag and /rerank endpoint exposure. Evidence is limited, and further verification is recommended to ensure accurate assessment and effective mitigation of the vulnerability. Defenders should verify system configurations, review compensating controls, and monitor for unusual activity to minimize potential impact. The CVE record indicates a denial-of-service vulnerability in Llama.cpp when started with the --reranking flag, allowing remote attackers to cause a std::bad_alloc and HTTP 500 error via a negative top_n value in a POST request to /rerank. Evidence is limited, and further verification is recommended. To

Technical summary

The Llama.cpp library, when started with the --reranking flag, is vulnerable to a denial-of-service attack. Remote attackers can cause a std::bad_alloc and HTTP 500 error by sending a POST request to /rerank with a negative top_n value. This issue impacts organizations using ggml's Llama.cpp, especially those exposing the /rerank endpoint to untrusted sources.

Defensive priority

Organizations using ggml's Llama.cpp should prioritize patching to prevent potential denial-of-service attacks.

Recommended defensive actions

• Apply patches or updates provided by the vendor to address the vulnerability
• Restrict access to the /rerank endpoint to trusted sources
• Monitor for unusual traffic patterns or errors related to the --reranking flag

Evidence notes

The CVE record indicates a denial-of-service vulnerability in Llama.cpp when started with the --reranking flag, allowing remote attackers to cause a std::bad_alloc and HTTP 500 error via a negative top_n value in a POST request to /rerank. Evidence is limited, and further verification is recommended.

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.