CVE-2026-133710 json
llama.cpp b11003 before b12556 Heap Out-of-Bounds Write via hrm_text Model Loader
published yesterday modified yesterday also EUVD-2026-131537
- cvss
- 8.5
- version 4.0, CNA
- cisa kev
- Not listed
- not in CISA's catalog
- epss
- 0.3%
- EPSS 0.3% chance of exploitation in 30 days, percentile 24%
- public exploits
- 0
- none known
Summary
llama.cpp b11003 before b12556 contains an integer overflow in the hrm_text loader load_arch_hparams() that allows attackers to bypass the block count check using crafted GGUF metadata. Attackers can supply a GGUF model whose layer factors wrap modulo 2^32, causing load_arch_tensors() to write tensor pointers past model.layers, corrupting heap memory and potentially enabling code execution.
Patch / fix
- ggml-org llama.cpp b12556
Is CVE-2026-133710 in your stack? Fencer shows which of your repositories are affected as soon as an advisory lands, and opens the fix as a pull request.
Check with FencerCISA SSVC
- Exploitation
- none
- Automatable
- no
- Technical impact
- total
- Assessed
- yesterday
Exploit prediction
0.31% chance of exploitation in the next 30 days, higher than 24% of all scored CVEs (FIRST EPSS, 2026-11-21).
CVSS scores
| assigned by | version | score | vector |
|---|---|---|---|
| CNA | 4.0 | 8.5 | CVSS:4.0/AV:L/AC:L/AT:N/PR:N/UI:P/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N |
| CNA | 3.1 | 7.8 | CVSS:3.1/AV:L/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H |
Affected products
- ggml-org llama.cpp
Classification
- Weakness
-
- CWE-190 Integer Overflow or Wraparound
References
-
github.com
/ggml-org/llama.cpp/pull/32601
issue-tracking, patch
-
github.com
/ggml-org/llama.cpp/commit/50518b64c6f2191a116d2760042f106c4b3387d7
patch
-
github.com
/ggml-org/llama.cpp/blob/b12555/src/models/hrm-text.cpp
technical-description
-
github.com
/ggml-org/llama.cpp/releases/tag/b12556
release-notes
-
vulncheck.com
/advisories/llama-cpp-b11003-before-b12556-heap-out-of-bounds-write-via-hrm-text-model-loader
third-party-advisory
Sources
- CVE List CVE-2026-133710
- NVD CVE-2026-133710
Fencer Intel keeps one entry per vulnerability: the CVE id wins, then GHSA, and records that alias each other are merged, except that two CVEs always keep separate entries. The same data is available as JSON.