CVE-2026-134540 json
ggml-org llama.cpp llama-server sampling.cpp common_sampler_init input validation
published 4h ago modified 4h ago also EUVD-2026-132566
- cvss
- 6.9
- version 4.0, CNA
- cisa kev
- Not listed
- not in CISA's catalog
- epss
- —
- no EPSS score yet
- public exploits
- 0
- none known
Summary
A security vulnerability has been detected in ggml-org llama.cpp 0.7.0-dev. This affects the function common_sampler_init of the file common/sampling.cpp of the component llama-server. Such manipulation of the argument grammar_triggers[].type leads to improper input validation. It is possible to launch the attack remotely. The exploit has been disclosed publicly and may be used. Upgrading to version b12405 mitigates this issue. The name of the patch is 9f26fe11c73c7e74a5e99e7f1b6370ad8e1cb22d. It is advisable to upgrade the affected component.
Patch / fix
- ggml-org llama.cpp b12405
Is CVE-2026-134540 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 FencerExploits and checks
- gist.github.com https://gist.github.com/Yunzez/90b17f91d7f1955805e0045a1a76fb6d#file-repro_llamacpp_grammar_triggers_abort-py referenced PoC, tagged "exploit" by a source, unverified
CVSS scores
| assigned by | version | score | vector |
|---|---|---|---|
| CNA | 4.0 | 6.9 | CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:N/VI:N/VA:L/SC:N/SI:N/SA:N/E:P |
| CNA | 3.1 | 5.3 | CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:L/E:P/RL:O/RC:C |
| CNA | 3.0 | 5.3 | CVSS:3.0/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:L/E:P/RL:O/RC:C |
| CNA | 2.0 | 5.0 | AV:N/AC:L/Au:N/C:N/I:N/A:P/E:POC/RL:OF/RC:C |
| NVD | 3.1 | 5.3 | CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:L |
Affected products
- ggml-org llama.cpp
Classification
- Weakness
-
- CWE-20 Improper Input Validation
References
-
github.com
/ggml-org/llama.cpp/pull/31478
issue-tracking, patch
-
github.com
/ggml-org/llama.cpp/commit/9f26fe11c73c7e74a5e99e7f1b6370ad8e1cb22d
patch
-
github.com
/ggml-org/llama.cpp/releases/tag/b12405
patch
-
gist.github.com
/Yunzez/90b17f91d7f1955805e0045a1a76fb6d#file-repro_llamacpp_grammar_triggers_abort-py
exploit
-
vuldb.com
/vuln/441681
vdb-entry, technical-description
-
vuldb.com
/vuln/441681/cti
signature, permissions-required
-
vuldb.com
/cve/CVE-2026-134540
third-party-advisory
-
vuldb.com
/submit/1052496
third-party-advisory
-
github.com
/ggml-org/llama.cpp/issues/30835
issue-tracking
-
github.com
/ggml-org/llama.cpp/
product
Sources
- CVE List CVE-2026-134540
- NVD CVE-2026-134540
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.