diff --git a/misc/llama-cpp/Makefile b/misc/llama-cpp/Makefile
index e880ababbdc1..0129909c8343 100644
--- a/misc/llama-cpp/Makefile
+++ b/misc/llama-cpp/Makefile
@@ -1,81 +1,81 @@
 PORTNAME=	llama-cpp
 DISTVERSIONPREFIX=	b
-DISTVERSION=	5371
+DISTVERSION=	5689
 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
 BROKEN_i386=	compilation fails, see https://github.com/ggerganov/llama.cpp/issues/9545
 
 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=	GGML_NATIVE \
 		FREEBSD_ALLOW_ADVANCED_CPU_FEATURES \
 		LLAMA_BUILD_TESTS
 CMAKE_TESTING_ON=	LLAMA_BUILD_TESTS
 
 # user for llama-server, only used when EXAMPLES=ON
 USER=		nobody
 SUB_LIST=	USER=${USER}
 
 OPTIONS_DEFINE=		CURL EXAMPLES VULKAN
 OPTIONS_DEFAULT=	CURL VULKAN
 OPTIONS_SUB=		yes
 
 CURL_DESCR=		Use libcurl to download model from an URL
 CURL_CMAKE_BOOL=	LLAMA_CURL
 CURL_USES=		localbase
 CURL_LIB_DEPENDS=	libcurl.so:ftp/curl
 
 EXAMPLES_CMAKE_BOOL=	LLAMA_BUILD_EXAMPLES
 
 VULKAN_DESC=		Vulkan GPU offload support
 VULKAN_CMAKE_BOOL=	GGML_VULKAN
 VULKAN_BUILD_DEPENDS=	glslc:graphics/shaderc \
 			vulkan-headers>0:graphics/vulkan-headers
 VULKAN_LIB_DEPENDS=	libvulkan.so:graphics/vulkan-loader
 
 BINARY_ALIAS=	git=false \
 		python=${PYTHON_CMD} # for tests
 
 post-patch: # set version in the code
 	@${REINPLACE_CMD} \
 		-e "s|set(BUILD_NUMBER 0)|set(BUILD_NUMBER ${DISTVERSION})|" \
 		${WRKSRC}/cmake/build-info.cmake
 
 do-test-ci: # build of tests fails, see https://github.com/ggerganov/llama.cpp/issues/10955
 	@cd ${WRKSRC} && \
 		${SETENV} ${MAKE_ENV} bash ci/run.sh ./tmp/results ./tmp/mnt
 
 .include <bsd.port.options.mk>
 
 .if ${PORT_OPTIONS:MEXAMPLES}
 USE_RC_SUBR=	llama-server
 .endif
 
 # tests as of 4458: 97% tests passed, 1 tests failed out of 31, see https://github.com/ggerganov/llama.cpp/issues/11036
 
 # tests as of 4649:
 # 88% tests passed, 4 tests failed out of 32
 # The following tests FAILED:
 #         18 - test-chat (Subprocess aborted)                    main   # see https://github.com/ggerganov/llama.cpp/issues/11705
 #         24 - test-gguf (SEGFAULT)                              main
 #         25 - test-backend-ops (SEGFAULT)                       main
 #         32 - test-eval-callback (SEGFAULT)                     curl eval-callback
 
 .include <bsd.port.mk>
diff --git a/misc/llama-cpp/distinfo b/misc/llama-cpp/distinfo
index 065e1989894b..c810b3383e6a 100644
--- a/misc/llama-cpp/distinfo
+++ b/misc/llama-cpp/distinfo
@@ -1,5 +1,5 @@
-TIMESTAMP = 1747201270
-SHA256 (ggerganov-llama.cpp-b5371_GH0.tar.gz) = cce50220507565b78423fc45a1c534dc088289ab898517a379fdbf733ffd72bf
-SIZE (ggerganov-llama.cpp-b5371_GH0.tar.gz) = 21147325
+TIMESTAMP = 1750213757
+SHA256 (ggerganov-llama.cpp-b5689_GH0.tar.gz) = 1ab18411c1f9885770b331bcc1f70ce9b0ad29f8591d7cf516ab6cdc6a6c1a63
+SIZE (ggerganov-llama.cpp-b5689_GH0.tar.gz) = 24994019
 SHA256 (nomic-ai-kompute-4565194_GH0.tar.gz) = 95b52d2f0514c5201c7838348a9c3c9e60902ea3c6c9aa862193a212150b2bfc
 SIZE (nomic-ai-kompute-4565194_GH0.tar.gz) = 13540496
diff --git a/misc/llama-cpp/pkg-plist b/misc/llama-cpp/pkg-plist
index 5b36c5809aaa..0a36dc3db062 100644
--- a/misc/llama-cpp/pkg-plist
+++ b/misc/llama-cpp/pkg-plist
@@ -1,66 +1,67 @@
 bin/convert_hf_to_gguf.py
 %%EXAMPLES%%bin/llama-batched
 %%EXAMPLES%%bin/llama-batched-bench
 %%EXAMPLES%%bin/llama-bench
 %%EXAMPLES%%bin/llama-cli
 %%EXAMPLES%%bin/llama-convert-llama2c-to-ggml
 %%EXAMPLES%%bin/llama-cvector-generator
 %%EXAMPLES%%bin/llama-embedding
 %%EXAMPLES%%bin/llama-eval-callback
 %%EXAMPLES%%bin/llama-export-lora
 %%EXAMPLES%%bin/llama-finetune
 %%EXAMPLES%%bin/llama-gen-docs
 %%EXAMPLES%%bin/llama-gguf
 %%EXAMPLES%%bin/llama-gguf-hash
 %%EXAMPLES%%bin/llama-gguf-split
 %%EXAMPLES%%bin/llama-gritlm
 %%EXAMPLES%%bin/llama-imatrix
 %%EXAMPLES%%bin/llama-lookahead
 %%EXAMPLES%%bin/llama-lookup
 %%EXAMPLES%%bin/llama-lookup-create
 %%EXAMPLES%%bin/llama-lookup-merge
 %%EXAMPLES%%bin/llama-lookup-stats
 %%EXAMPLES%%bin/llama-mtmd-cli
 %%EXAMPLES%%bin/llama-parallel
 %%EXAMPLES%%bin/llama-passkey
 %%EXAMPLES%%bin/llama-perplexity
 %%EXAMPLES%%bin/llama-quantize
 %%EXAMPLES%%bin/llama-retrieval
 %%EXAMPLES%%bin/llama-run
 %%EXAMPLES%%bin/llama-save-load-state
 %%EXAMPLES%%bin/llama-server
 %%EXAMPLES%%bin/llama-simple
 %%EXAMPLES%%bin/llama-simple-chat
 %%EXAMPLES%%bin/llama-speculative
 %%EXAMPLES%%bin/llama-speculative-simple
 %%EXAMPLES%%bin/llama-tokenize
 %%EXAMPLES%%bin/llama-tts
-%%VULKAN%%bin/vulkan-shaders-gen
 include/ggml-alloc.h
 include/ggml-backend.h
 include/ggml-blas.h
 include/ggml-cann.h
 include/ggml-cpp.h
 include/ggml-cpu.h
 include/ggml-cuda.h
 include/ggml-kompute.h
 include/ggml-metal.h
 include/ggml-opt.h
 include/ggml-rpc.h
 include/ggml-sycl.h
 include/ggml-vulkan.h
 include/ggml.h
 include/gguf.h
 include/llama-cpp.h
 include/llama.h
+include/mtmd-helper.h
+include/mtmd.h
 lib/cmake/ggml/ggml-config.cmake
 lib/cmake/ggml/ggml-version.cmake
 lib/cmake/llama/llama-config.cmake
 lib/cmake/llama/llama-version.cmake
 lib/libggml-base.so
 lib/libggml-cpu.so
 %%VULKAN%%lib/libggml-vulkan.so
 lib/libggml.so
 lib/libllama.so
-lib/libmtmd_shared.so
+lib/libmtmd.so
 libdata/pkgconfig/llama.pc