CVE-2026-116840: llama.cpp b5854 before b11275 Integer Overflow in gguf_init_from_reader
llama.cpp b5854 before b11275 contains an integer overflow vulnerability in ggufinitfromreader that bypasses the CVE-2025-53630 size check when padding near-SIZEMAX tensor sizes. Attackers can supply a crafted GGUF file whose tensor size pads to zero, causing an undersized data blob and out-of-bounds reads by consumers.
Affected Software
Event History
Frequently Asked Questions
Which deployments are exposed?
Deployments using llama.cpp builds from b5854 up to, but not including, b11275 are affected when they process a crafted GGUF file. The issue is triggered in gguf_init_from_reader.
What does exploitation require?
An attacker needs to cause a user to process a crafted GGUF file with tensor sizes that pad near SIZE_MAX. No privileges are required, but user interaction is required.
How can I determine whether an installation is affected?
Check the llama.cpp build version. Builds before b11275 are affected within the stated range beginning at b5854; b11275 is outside the affected range.