Vulnerability Description
llama.cpp is an inference of several LLM models in C/C++. Prior to version b5721, there is a signed vs. unsigned integer overflow in llama.cpp's tokenizer implementation (llama_vocab::tokenize) (src/llama-vocab.cpp:3036) resulting in unintended behavior in tokens copying size comparison. Allowing heap-overflowing llama.cpp inferencing engine with carefully manipulated text input during tokenization process. This issue has been patched in version b5721.
CVSS Score
HIGH
Affected Products
| Vendor | Product | Versions |
|---|---|---|
| Ggml | Llama.Cpp | < b5721 |
Related Weaknesses (CWE)
References
- https://github.com/ggml-org/llama.cpp/commit/dd6e6d0b6a4bbe3ebfc931d1eb14db2f2b1Patch
- https://github.com/ggml-org/llama.cpp/security/advisories/GHSA-7rxv-5jhh-j6xxExploitVendor Advisory
- https://github.com/ggml-org/llama.cpp/security/advisories/GHSA-7rxv-5jhh-j6xxExploitVendor Advisory
FAQ
What is CVE-2025-52566?
CVE-2025-52566 is a vulnerability with a CVSS score of 8.6 (HIGH). llama.cpp is an inference of several LLM models in C/C++. Prior to version b5721, there is a signed vs. unsigned integer overflow in llama.cpp's tokenizer implementation (llama_vocab::tokenize) (src/l...
How severe is CVE-2025-52566?
CVE-2025-52566 has been rated HIGH with a CVSS base score of 8.6/10. Review the CVSS metrics above for detailed severity breakdown.
Is there a patch for CVE-2025-52566?
Check the references section above for vendor advisories and patch information. Affected products include: Ggml Llama.Cpp.