Ggml provides a machine-learning inference library that, despite a narrow product focus, underpins widely deployed implementations such as llama.cpp and has achieved prominent adoption in the generative AI ecosystem. Vulnerabilities affecting the vendor skew strongly toward critical-severity outcomes and recur through memory-safety weakness classes—out-of-bounds writes and reads, integer overflows, heap-based buffer overflows, and improper bounds restrictions—that are characteristic of performance-critical native code handling untrusted model inputs and tensor operations. Defenders should track this vendor's releases closely given the increasing deployment of these inference engines in production systems; live severity, exploitation, and exposure counts are shown alongside this summary.
The number and severity of CVEs published that impact products developed by Ggml over time
Signals from CVEs in this vendor scope (16 CVEs).
16 CVEs · Highest risk first
| CVE | Published | CVSS | Risk | KEV | Exploit |
|---|---|---|---|---|---|
CVE-2026-34159CRITICAL llama.cpp is an inference of several LLM models in C/C++. Prior to version b8492, the RPC backend's deserialize_tensor() skips all bounds validation when a tensor's buffer field is | Apr 1, 2026 | 9.8 | 36 | NO | NO |
CVE-2024-42479CRITICAL llama.cpp provides LLM inference in C/C++. The unsafe `data` pointer member in the `rpc_tensor` structure can cause arbitrary address writing. This vulnerability is fixed in b3561. | Aug 12, 2024 | 9.8 | 33 | NO | NO |
CVE-2026-21869CRITICAL llama.cpp is an inference of several LLM models in C/C++. In commits 55d4206c8 and prior, the n_discard parameter is parsed directly from JSON input in the llama.cpp server's compl | Jan 8, 2026 | 9.8 | 32 | NO | NO |
CVE-2024-42478CRITICAL llama.cpp provides LLM inference in C/C++. The unsafe `data` pointer member in the `rpc_tensor` structure can cause arbitrary address reading. This vulnerability is fixed in b3561. | Aug 12, 2024 | 9.8 | 29 | NO | NO |
CVE-2024-21836CRITICAL A heap-based buffer overflow vulnerability exists in the GGUF library header.n_tensors functionality of llama.cpp Commit 18c2e17. A specially crafted .gguf file can lead to code ex | Feb 26, 2024 | 9.8 | 28 | NO | NO |
CVE-2026-33298HIGH llama.cpp is an inference of several LLM models in C/C++. Prior to b7824, an integer overflow vulnerability in the `ggml_nbytes` function allows an attacker to bypass memory valida | Mar 24, 2026 | 7.8 | 27 | NO | NO |
CVE-2024-21825CRITICAL A heap-based buffer overflow vulnerability exists in the GGUF library GGUF_TYPE_ARRAY/GGUF_TYPE_STRING parsing functionality of llama.cpp Commit 18c2e17. A specially crafted .gguf | Feb 26, 2024 | 9.8 | 27 | NO | NO |
CVE-2024-21802CRITICAL A heap-based buffer overflow vulnerability exists in the GGUF library info->ne functionality of llama.cpp Commit 18c2e17. A specially crafted .gguf file can lead to code executi | Feb 26, 2024 | 9.8 | 27 | NO | NO |
CVE-2026-27940HIGH llama.cpp is an inference of several LLM models in C/C++. Prior to b8146, the gguf_init_from_file_impl() in gguf.cpp is vulnerable to an Integer overflow, leading to an undersized | Mar 12, 2026 | 7.8 | 26 | NO | NO |
CVE-2024-23605CRITICAL A heap-based buffer overflow vulnerability exists in the GGUF library header.n_kv functionality of llama.cpp Commit 18c2e17. A specially crafted .gguf file can lead to code executi | Feb 26, 2024 | 9.8 | 26 | NO | NO |
Signals from CVEs in this vendor scope (16 CVEs).
An overview of all social media posts that mention a CVE ID that affects a product developed by Ggml.
Media articles that mention a CVE ID that affects a product developed by Ggml — matched by CVE ID, not by vendor name.