CVE-2025-49847
llama.cpp Vulnerable to Buffer Overflow via Malicious GGUF Model
Published Jun 17, 2025
CVSS 3.1 · 8.8 High
EPSS < 1%
KEV · Not listed
Description
llama.cpp is an inference of several LLM models in C/C++. Prior to version b5662, an attacker‐supplied GGUF model vocabulary can trigger a buffer overflow in llama.cpp’s vocabulary‐loading code. Specifically, the helper _try_copy in llama.cpp/src/vocab.cpp: llama_vocab::impl::token_to_piece() casts a very large size_t token length into an int32_t, causing the length check (if (length < (int32_t)size)) to be bypassed. As a result, memcpy is still called with that oversized size, letting a malicious model overwrite memory beyond the intended buffer. This can lead to arbitrary memory corruption and potential code execution. This issue has been patched in version b5662.
Affected Products
-
CVE record
-
NVD configurations
-
Red Hat packages
No data
-
OpenCVE enrichment
No data
Default status is the baseline for the product, each version can override it (e.g. patched versions marked unaffected).
| Vendor | Product | Default status | Versions | ||||||
|---|---|---|---|---|---|---|---|---|---|
| ggml-org | llama.cpp | affected |
|
No data.
No data available yet.
Remediation
Tracking
Advisories
| Source | ID | Title |
|---|---|---|
 EUVD |
EUVD-2025-18632 | llama.cpp is an inference of several LLM models in C/C++. Prior to version b5662, an attacker‐supplied GGUF model vocabulary can trigger a buffer overflow in llama.cpp’s vocabulary‐loading code. Specifically, the helper _try_copy in llama.cpp/src/vocab.cpp: llama_vocab::impl::token_to_piece() casts a very large size_t token length into an int32_t, causing the length check (if (length < (int32_t)size)) to be bypassed. As a result, memcpy is still called with that oversized size, letting a malicious model overwrite memory beyond the intended buffer. This can lead to arbitrary memory corruption and potential code execution. This issue has been patched in version b5662. |
Metrics
No CVSS v4.0
Attack Vector Network
Attack Complexity Low
Privileges Required None
Scope Unchanged
Confidentiality Impact High
Integrity Impact High
Availability Impact High
User Interaction Required
No CVSS v3.0
No CVSS v2
This CVE is not in the KEV list.
Exploit prediction score: < 1% (0.553% probability within 30 days)
Exploitation poc
Automatable no
Technical Impact total
References
History
Aug 27, 2025 2:00 PM
First appearance in the feed. Affected versions updated. Affected vendors updated
View changes
| Type | Values Removed | Values Added |
|---|---|---|
| First Time appeared |
Ggml
Ggml llama.cpp |
|
| CPEs | cpe:2.3:a:ggml:llama.cpp:*:*:*:*:*:*:*:* | |
| Vendors & Products |
Ggml
Ggml llama.cpp |
Jun 18, 2025 2:15 PM
Scoring metrics updated
View changes
| Type | Values Removed | Values Added |
|---|---|---|
| Metrics |
ssvc
|
Jun 17, 2025 8:15 PM
Title updated
View changes
| Type | Values Removed | Values Added |
|---|---|---|
| Title | llama.cpp Vulnerable to Buffer Overflow via Malicious GGUF Model |
Jun 17, 2025 8:45 PM
Description updated. Weaknesses updated. Added references. Scoring metrics updated
View changes
| Type | Values Removed | Values Added |
|---|---|---|
| Description | llama.cpp is an inference of several LLM models in C/C++. Prior to version b5662, an attacker‐supplied GGUF model vocabulary can trigger a buffer overflow in llama.cpp’s vocabulary‐loading code. Specifically, the helper _try_copy in llama.cpp/src/vocab.cpp: llama_vocab::impl::token_to_piece() casts a very large size_t token length into an int32_t, causing the length check (if (length < (int32_t)size)) to be bypassed. As a result, memcpy is still called with that oversized size, letting a malicious model overwrite memory beyond the intended buffer. This can lead to arbitrary memory corruption and potential code execution. This issue has been patched in version b5662. | |
| Weaknesses | CWE-119 CWE-195 |
|
| References |
| |
| Metrics |
cvssV3_1
|
