CVE-2026-43631

Name of the Vulnerable Software and Affected Versions llama.cpp versions b7492 through b9060

Description A use-after-free issue exists in the llama-server when the --sleep-idle-seconds feature is enabled. This occurs when the server transitions to sleep mode, causing concurrent worker threads to dereference a freed vocab pointer. An unauthenticated remote attacker can exploit this by sending requests to affected endpoints, potentially reclaiming the freed pointer with controlled data to achieve remote code execution.

Recommendations Update llama.cpp to a version later than b9060. As a temporary mitigation, disable the --sleep-idle-seconds feature.