Back to CVE records
High 7.8Published Aug 6, 2026

CVE-2026-43628llama.cpp b3978–b9058 Integer Underflow via DRY Sampler

Updated under an hour ago ()

Does this affect you?

It affects ggml-org llama.cpp and 1 other product. Add what you run to find out whether that includes you.

No exploitation reported and no EPSS estimate yet.

llama.cpp builds b3978 through b9058 contain an integer underflow and out-of-bounds read vulnerability in the DRY sampler that allows unauthenticated attackers to trigger a heap buffer underflow by sending a crafted HTTP request with dry_allowed_length set to INT32_MIN to the /v1/completions or /v1/chat/completions endpoints. Attackers can exploit this vulnerability to crash the server with SIGSEGV causing denial of service for all connected users, or corrupt token sampling probabilities by reading garbage values from memory before the allocated buffer.

CVSS
7.8 HIGH (v3.1, CNA)CVSS:3.1/AV:L/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H
Weakness (CWE)
CWE-191, CWE-125
Known exploited
Not in CISA's KEV catalog
EPSS (exploitation in the next 30 days)
Not loaded for this CVE
Assigned by
VulnCheck
NVD analysis
Analyzed
Record last updated
Aug 14, 2026
First seen by InfoSec Signals
Oct 9, 2026

Affected products

Affected products
ggml-org llama.cpp
ggml_org:llama.cpp (from the name)
from b3978 through b9058; from 0.11.0 through 0.17.1
CNA record
ggml llama.cpp
ggml:llama.cpp
from b3978 through b9058
NVD

References

  • mitigation, patch, third-party-advisory
  • third-party-advisory, patch

The CVE record is authoritative. Products marked “from the name” were identified from the CNA's free-text names, not a CPE, and are matched less certainly. This product uses the NVD API but is not endorsed or certified by the NVD.

Your cookie choices
We use essential cookies to run this site, and, only with your consent, an advertising cookie from Google to measure whether our ads lead to sign-ups and subscriptions. See our for details.