diff --git a/backends/llamacpp/CMakeLists.txt b/backends/llamacpp/CMakeLists.txt
index 890d99da..9f08d0f3 100644
--- a/backends/llamacpp/CMakeLists.txt
+++ b/backends/llamacpp/CMakeLists.txt
@@ -6,6 +6,7 @@ set(CMAKE_CXX_STANDARD 23)
 include(FetchContent)
 
 set(LLAMA_CPP_TARGET_VERSION "b3837" CACHE STRING "Version of llama.cpp to build against")
+set(LLAMA_BUILD_COMMON ON)
 set(LLAMA_CPP_TARGET_CUDA_ARCHS "75-real;80-real;86-real;89-real;90-real" CACHE STRING "CUDA arch(s) to build")
 option(LLAMA_CPP_BUILD_OFFLINE_RUNNER "Flag to build the standalone c++ backend runner")
 option(LLAMA_CPP_BUILD_CUDA "Flag to build CUDA enabled inference through llama.cpp")
@@ -28,7 +29,7 @@ fetchcontent_declare(
         llama
         #    DOWNLOAD_EXTRACT_TIMESTAMP TRUE
         GIT_REPOSITORY https://github.com/ggerganov/llama.cpp.git
-        GIT_TAG b3837
+        GIT_TAG b3958
         GIT_SHALLOW FALSE
 )
 
@@ -41,7 +42,8 @@ target_link_libraries(tgi_llama_cpp_backend_impl PUBLIC fmt::fmt spdlog::spdlog
 if (${LLAMA_CPP_BUILD_OFFLINE_RUNNER})
     message(STATUS "Building llama.cpp offline runner")
     add_executable(tgi_llama_cpp_offline_runner offline/main.cpp)
-    target_link_libraries(tgi_llama_cpp_offline_runner tgi_llama_cpp_backend_impl)
+
+    target_link_libraries(tgi_llama_cpp_offline_runner PUBLIC tgi_llama_cpp_backend_impl llama common)
 endif ()