# cve.now: Ggml-org

The newest entries naming Ggml-org as an affected vendor (up to 200), of 40 in all, 0 known exploited. Older ones are in the full catalog at https://cve.now/latest.md.

- [CVE-2026-119052](https://cve.now/CVE-2026-119052.md): llama.cpp b2877 before b11935 ggml-rpc Server DoS via Oversized recv\_msg Length (CVSS 8.7)
- [CVE-2026-118375](https://cve.now/CVE-2026-118375.md): llama.cpp b8841 before b11865 Stack-based Buffer Overflow via ggml-rpc RDMA rdma\_recv() (CVSS 9.2)
- [CVE-2026-116840](https://cve.now/CVE-2026-116840.md): llama.cpp b5854 before b11275 Integer Overflow in gguf\_init\_from\_reader (CVSS 6.9)
- [CVE-2026-116829](https://cve.now/CVE-2026-116829.md): llama.cpp b10398 before b11301 Integer Overflow in gguf\_init\_from\_reader (CVSS 6.9)
- [CVE-2026-116816](https://cve.now/CVE-2026-116816.md): llama.cpp b5899 before b10261 Heap Out-of-Bounds Write via plamo2 Tokenizer (CVSS 8.5)
- [CVE-2025-71439](https://cve.now/CVE-2025-71439.md): llama.cpp before b5649 Heap Out-of-Bounds Read via llama\_vocab::impl::print\_info (CVSS 6.8)
- [CVE-2024-58400](https://cve.now/CVE-2024-58400.md): llama.cpp before b3701 Out-of-Bounds Read in llama\_decode via Token IDs (CVSS 8.8)
- [CVE-2024-58399](https://cve.now/CVE-2024-58399.md): llama.cpp before b2731 Out-of-Bounds Read via GGUF Tensor Offsets (CVSS 6.9)
- [CVE-2026-116467](https://cve.now/CVE-2026-116467.md): ggml-org llama.cpp Mirostat Sampler llama-sampler.cpp assertion (CVSS 6.9)
- [CVE-2026-116096](https://cve.now/CVE-2026-116096.md): ggml-org llama.cpp Pocket-TTS pockettts-gen.cpp list\_pockettts\_state\_slots out-of-bounds write (CVSS 4.8)
- [CVE-2026-110810](https://cve.now/CVE-2026-110810.md): ggml-org whisper.cpp Model Loader whisper\_model\_load out-of-bounds (CVSS 2.4)
- [CVE-2026-107183](https://cve.now/CVE-2026-107183.md): llama.cpp before b11393 Use-After-Free via common\_chat\_peg\_mapper chat\_parser (CVSS 9.2)
- [CVE-2026-86317](https://cve.now/CVE-2026-86317.md): ggml-org llama.cpp RPC Server ggml-rpc.cpp deserialize\_tensor assertion (CVSS 6.9)
- [CVE-2026-78148](https://cve.now/CVE-2026-78148.md): ggml-org llama.cpp ggml-RPC Server ggml-rpc.cpp graph\_compute null pointer dereference (CVSS 6.9)
- [CVE-2026-78147](https://cve.now/CVE-2026-78147.md): ggml-org llama.cpp ggml-RPC Server ggml-rpc.cpp deserialize\_tensor deserialization (CVSS 6.9)
- [CVE-2026-43632](https://cve.now/CVE-2026-43632.md): llama.cpp b7492–b9060 Use-After-Free in Tokenization Endpoints (CVSS 9.2)
- [CVE-2026-43631](https://cve.now/CVE-2026-43631.md): llama.cpp b7492–b9060 Use-After-Free RCE via llama-server (CVSS 9.2)
- [CVE-2026-43630](https://cve.now/CVE-2026-43630.md): llama.cpp b4882-b8487 Out-of-Bounds Read Information Disclosure (CVSS 6.3)
- [CVE-2026-43629](https://cve.now/CVE-2026-43629.md): llama.cpp b4882–b9058 Buffer Overflow in KV Cache State Restore (CVSS 9.2)
- [CVE-2026-43628](https://cve.now/CVE-2026-43628.md): llama.cpp b3978–b9058 Integer Underflow via DRY Sampler (CVSS 8.5)
- [CVE-2026-43627](https://cve.now/CVE-2026-43627.md): llama.cpp b1283–b9058 Integer Overflow in llama\_batch\_init() Function (CVSS 8.5)
- [CVE-2026-70640](https://cve.now/CVE-2026-70640.md): llama.cpp b1886–b7445 Race Condition Use-After-Free via llama-android.cpp (CVSS 7.3)
- [CVE-2026-70639](https://cve.now/CVE-2026-70639.md): llama.cpp b1886–b7445 Null Pointer Dereference DoS via llama-android.cpp (CVSS 6.8)
- [CVE-2026-70638](https://cve.now/CVE-2026-70638.md): llama.cpp b1886–b7445 Integer Overflow via new\_1batch() in llama-android.cpp (CVSS 8.5)
- [CVE-2026-43622](https://cve.now/CVE-2026-43622.md): llama.cpp b1886–b7445 Double Free via llama-android.cpp (CVSS 8.5)
- [CVE-2026-18581](https://cve.now/CVE-2026-18581.md): ggml-org llama.cpp Jinja Minja Template parser.cpp assertion (CVSS 4.8)
- [CVE-2026-17513](https://cve.now/CVE-2026-17513.md): ggml-org whisper.cpp ggml.c ggml\_ftype\_to\_ggml\_type assertion (CVSS 4.8)
- [CVE-2026-17512](https://cve.now/CVE-2026-17512.md): ggml-org whisper.cpp log\_mel\_spectrogram out-of-bounds (CVSS 4.8)
- [CVE-2026-17501](https://cve.now/CVE-2026-17501.md): ggml-org llama.cpp JSON-Schema-to-GBNF Conversion json-schema-to-grammar.cpp transform recursion (CVSS 6.9)
- [CVE-2026-17500](https://cve.now/CVE-2026-17500.md): ggml-org llama.cpp json-schema-to-grammar.cpp \_visit\_pattern null pointer dereference (CVSS 6.9)
- [CVE-2026-10298](https://cve.now/CVE-2026-10298.md): ggml-org whisper.cpp ggml.c whisper\_model\_load null pointer dereference (CVSS 4.8)
- [CVE-2026-34159](https://cve.now/CVE-2026-34159.md): llama.cpp: Unauthenticated RCE via GRAPH\_COMPUTE buffer=0 bypass in llama.cpp RPC backend (CVSS 9.8)
- [CVE-2026-33298](https://cve.now/CVE-2026-33298.md): llama.cpp has a Heap Buffer Overflow via Integer Overflow in GGUF Tensor Parsing (CVSS 7.8)
- [CVE-2026-27940](https://cve.now/CVE-2026-27940.md): llama.cpp has a Heap Buffer Overflow via Integer Overflow in \`mem\_size\` Calculation — Bypass of CVE-2025-53630 Fix (CVSS 7.8)
- [CVE-2026-2069](https://cve.now/CVE-2026-2069.md): ggml-org llama.cpp GBNF Grammar llama-grammar.cpp llama\_grammar\_advance\_stack stack-based overflow (CVSS 4.8)
- [CVE-2026-21869](https://cve.now/CVE-2026-21869.md): llama.cpp has Out-of-bounds Write in llama-server (CVSS 8.8)
- [CVE-2025-14569](https://cve.now/CVE-2025-14569.md): ggml-org whisper.cpp common-whisper.cpp read\_audio\_data use after free (CVSS 4.8)
- [CVE-2025-53630](https://cve.now/CVE-2025-53630.md): Integer Overflow in GGUF Parser can lead to Heap Out-of-Bounds Read/Write in gguf (CVSS 8.9)
- [CVE-2025-52566](https://cve.now/CVE-2025-52566.md): llama.cpp tokenizer signed vs. unsigned heap overflow (CVSS 8.6)
- [CVE-2025-49847](https://cve.now/CVE-2025-49847.md): llama.cpp Vulnerable to Buffer Overflow via Malicious GGUF Model (CVSS 8.8)

---

Source: cve.now
Built by [Sazabi](https://www.sazabi.com/?utm_source=cve-now&utm_medium=markdown), AI-native observability and security: automatic issue detection, plain-language log search and live dashboards.
Page: https://cve.now/vendors/ggml-org
Site guide for agents: https://cve.now/llms.txt
Uses NVD data; not endorsed by NVD.