diff --git a/gpt4all-backend/llama.cpp-mainline b/gpt4all-backend/llama.cpp-mainline
index d3915bb5..bebf55c9 160000
--- a/gpt4all-backend/llama.cpp-mainline
+++ b/gpt4all-backend/llama.cpp-mainline
@@ -1 +1 @@
-Subproject commit d3915bb50ed8872242edf798f3d0518028569650
+Subproject commit bebf55c962faa19ff11e030381cf72870bb515d6
diff --git a/gpt4all-backend/llama.cpp.cmake b/gpt4all-backend/llama.cpp.cmake
index 03e01c48..d8ec4d10 100644
--- a/gpt4all-backend/llama.cpp.cmake
+++ b/gpt4all-backend/llama.cpp.cmake
@@ -236,7 +236,9 @@ if (LLAMA_KOMPUTE)
           kompute/op_rmsnorm.comp
           kompute/op_diagmask.comp
           kompute/op_mul_mat_f16.comp
+          kompute/op_mul_mat_mat_f16.comp
           kompute/op_mul_mat_q4_0.comp
+          kompute/op_mul_mat_mat_q4_0.comp
           kompute/op_mul_mat_q4_1.comp
           kompute/op_getrows_f16.comp
           kompute/op_getrows_q4_0.comp
@@ -263,7 +265,9 @@ if (LLAMA_KOMPUTE)
           shaderop_rmsnorm.h
           shaderop_diagmask.h
           shaderop_mul_mat_f16.h
+          shaderop_mul_mat_mat_f16.h
           shaderop_mul_mat_q4_0.h
+          shaderop_mul_mat_mat_q4_0.h
           shaderop_mul_mat_q4_1.h
           shaderop_getrows_f16.h
           shaderop_getrows_q4_0.h