llama.cpp: Llama.cpp builds b3978 through b9058 contain an integer underflow and out-of-bounds read…
llama.cpp
More for llama.cpp RSS feed for llama.cpp
- Severity
- High, CVSS 7.8
- CVSS vector
- CVSS:3.1/AV:L/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H
- Record
- CVE-2026-43628
- Affects
- llama.cpp
- Known exploited
- Not listed in the CISA Known Exploited Vulnerabilities catalog
- Published
- Record first published by the source
- August 6, 2026
- Source last modified
- October 23, 2026
Description
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.
Description quoted from the NVD record.
Who is affected
This flaw affects llama.cpp, made by ggml, a program that runs AI language models on your own computers or servers so other apps can send it questions and get answers. It affects builds b3978 through b9058, in a feature that helps keep the AI's answers from repeating themselves. Someone who does not need to log in can send a specially crafted web request to the parts of the server that answer text and chat requests. This can crash the server, cutting off everyone using it, or make it read leftover data from memory that disturbs how it picks words for its answers. This flaw is rated High.
What to do
- The person who installed or runs llama.cpp for your organization, or the IT provider who manages it, should act on this flaw.
- Find out which build of llama.cpp you run. If it is any build from b3978 through b9058, it is affected.
- No fixed version is named for this flaw yet, so check the vendor advisory linked on this page for a fixed release and install it as soon as one is available.
- If an IT provider or vendor runs llama.cpp for you, send them this advisory and ask when the update will be done.
Analysis written from the source record as of October 23, 2026 and reviewed before publishing.