Vulnerability · Published Aug 6, 2026 · Assigned by VulnCheck

CVE-2026-43632llama.cpp b7492–b9060 Use-After-Free in Tokenization Endpoints

High 8.1

In plain English

A high-severity use-after-free in llama.cpp from ggml-org. There are no confirmed reports of exploitation so far. The chance it is exploited in the next 30 days is estimated at 0.5% (higher than 41% of all known vulnerabilities).

Severity (CVSS 3.1)8.1High

Exploit odds (30 days)0.5%top 58.7% of all CVEs

Exploited in the wildNo reportsCISA SSVC: none

Attack toolingNoneno public tooling known

Official description

llama.cpp builds b7492 through the latest b9060 contains a use-after-free vulnerability in llama-server affecting six tokenization endpoints (/tokenize, /detokenize, /infill, /apply-template, /rerank, and /anthropic/count_tokens) that bypass the task queue and access ctx_server.vocab directly on HTTP worker threads. Attackers can exploit a time-of-check-time-of-use race condition where the main thread destroys and frees vocab after the synchronization lock is released but before the handler finishes using it, causing a crash or potential code execution when --sleep-idle-seconds is configured.

Affected packages & fixed versions

Ecosystem	Package	Fixed in	Advisory
Debian:14	llama.cpp	no fix listed	DEBIAN-CVE-2026-43632
Ubuntu:26.04:LTS	llama.cpp	no fix listed	UBUNTU-CVE-2026-43632
Ubuntu:26.10	llama.cpp	no fix listed	UBUNTU-CVE-2026-43632

Affected products (as reported)

Vendor	Product	Versions
ggml-org	llama.cpp	b7492 – through b9060

Weakness: CWE-416, CWE-367 · Vector: CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:H/I:H/A:H

References

https://github.com/Vladimir-tokarev-cyera/llama-cpp-security-patches mitigation
https://www.vulncheck.com/advisories/llama-cpp-b7492-b9060-use-after-free-in-tokenization-endpoints third-party-advisory