diff --git a/misc/llama-cpp/Makefile b/misc/llama-cpp/Makefile
index 87b37524a792..96fc2fa9b0b8 100644
--- a/misc/llama-cpp/Makefile
+++ b/misc/llama-cpp/Makefile
@@ -1,40 +1,40 @@
 PORTNAME=	llama-cpp
 DISTVERSIONPREFIX=	b
-DISTVERSION=	2929
+DISTVERSION=	2940
 CATEGORIES=	misc # machine-learning
 
 MAINTAINER=	yuri@FreeBSD.org
 COMMENT=	Facebook's LLaMA model in C/C++ # '
 WWW=		https://github.com/ggerganov/llama.cpp
 
 LICENSE=	MIT
 LICENSE_FILE=	${WRKSRC}/LICENSE
 
 BROKEN_armv7=	clang crashes, see https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=278810
 
 USES=		cmake:testing compiler:c++11-lang python:run shebangfix
 USE_LDCONFIG=	yes
 
 USE_GITHUB=	yes
 GH_ACCOUNT=	ggerganov
 GH_PROJECT=	llama.cpp
 GH_TUPLE=	nomic-ai:kompute:4565194:kompute/kompute
 
 SHEBANG_GLOB=	*.py
 
 CMAKE_ON=	BUILD_SHARED_LIBS
 CMAKE_OFF=	LLAMA_BUILD_TESTS
 CMAKE_TESTING_ON=	LLAMA_BUILD_TESTS
 
 LDFLAGS+=	-pthread
 
 OPTIONS_DEFINE=		EXAMPLES
 OPTIONS_SUB=		yes
 
 EXAMPLES_CMAKE_BOOL=	LLAMA_BUILD_EXAMPLES
 
 BINARY_ALIAS=	git=false
 
 # 1 test fails due to a missing model file (stories260K.gguf)
 
 .include <bsd.port.mk>
diff --git a/misc/llama-cpp/distinfo b/misc/llama-cpp/distinfo
index 5b788e6de661..a5e8bde0e886 100644
--- a/misc/llama-cpp/distinfo
+++ b/misc/llama-cpp/distinfo
@@ -1,5 +1,5 @@
-TIMESTAMP = 1716103078
-SHA256 (ggerganov-llama.cpp-b2929_GH0.tar.gz) = 00a1e83286c53c497efa9324c9944f1bc0a2c695f19b2fc93977b080c7d8e5b0
-SIZE (ggerganov-llama.cpp-b2929_GH0.tar.gz) = 20130714
+TIMESTAMP = 1716178840
+SHA256 (ggerganov-llama.cpp-b2940_GH0.tar.gz) = 554155abf1b2ef8ebc3256e72329420ccdd5a5d2a0ec4d4e08ba6a99b2722a40
+SIZE (ggerganov-llama.cpp-b2940_GH0.tar.gz) = 20149015
 SHA256 (nomic-ai-kompute-4565194_GH0.tar.gz) = 95b52d2f0514c5201c7838348a9c3c9e60902ea3c6c9aa862193a212150b2bfc
 SIZE (nomic-ai-kompute-4565194_GH0.tar.gz) = 13540496