diff --git a/ggml-cuda.cu b/ggml-cuda.cu
index 48c7c83df..e31d494b2 100644
--- a/ggml-cuda.cu
+++ b/ggml-cuda.cu
@@ -2542,7 +2542,7 @@ bool ggml_cuda_is_gpu_offloading(struct ggml_tensor * tensor) {
         || (tensor->src1 != nullptr && tensor->src1->backend == GGML_BACKEND_GPU);
 }
 
-bool ggml_cuda_compute_forward(struct ggml_compute_params * params, struct ggml_tensor * tensor){
+bool ggml_cuda_compute_forward(const struct ggml_compute_params * params, struct ggml_tensor * tensor){
     ggml_cuda_func_t func;
     const bool any_on_device = ggml_cuda_is_gpu_offloading(tensor);
 
diff --git a/ggml-cuda.h b/ggml-cuda.h
index 70bd65e22..efb5bb38f 100644
--- a/ggml-cuda.h
+++ b/ggml-cuda.h
@@ -32,7 +32,7 @@ void   ggml_cuda_assign_buffers_no_scratch(struct ggml_tensor * tensor);
 void   ggml_cuda_set_main_device(int main_device);
 void   ggml_cuda_set_scratch_size(size_t scratch_size);
 void   ggml_cuda_free_scratch(void);
-bool   ggml_cuda_compute_forward(struct ggml_compute_params * params, struct ggml_tensor * tensor);
+bool   ggml_cuda_compute_forward(const struct ggml_compute_params * params, struct ggml_tensor * tensor);
 
 #ifdef  __cplusplus
 }