CVE-2026-70638

llama.cpp builds b1886 through b7445 contain an integer overflow vulnerability in the LLaMA-Android JNI wrapper where the new_1batch() function multiplies sizeof(llama_seq_id) by an attacker-controlled n_seq_max parameter without overflow validation, causing heap buffer allocation to wrap and allocate insufficient memory. Attackers can exploit this by providing a crafted n_seq_max value through a malicious model file or JNI call to trigger heap corruption and achieve denial of service or arbitrary code execution on Android applications using the LLaMA-Android binding.

Score

CVSS 4.0 8.5 High

Status

Analyzed - disclosure@vulncheck.com

Published

2026-08-06

Last modified

2026-10-24

Weaknesses

CWE-122, CWE-190

Affected products

• Llama.cpp (Ggml)

References

• https://github.com/ggml-org/llama.cpp/releases/tag/b7446
• https://github.com/ggml-org/llama.cpp/commit/5c0d18881e0e9794c96b2602736b758bac9d9388
• https://github.com/Vladimir-tokarev-cyera/llama-cpp-security-patches
• https://www.vulncheck.com/advisories/llama-cpp-b1886-b7445-integer-overflow-via-new-1batch-in-llama-android-cpp