diff --git a/src/llama-arch.cpp b/src/llama-arch.cpp
index c9eead18a..e7d027c98 100644
--- a/src/llama-arch.cpp
+++ b/src/llama-arch.cpp
@@ -767,8 +767,9 @@ static const std::map<llm_tensor, llm_tensor_info> LLM_TENSOR_INFOS = {
     {LLM_TENSOR_NEXTN_SHARED_HEAD_HEAD,     {LLM_TENSOR_LAYER_REPEATING, GGML_OP_MUL_MAT}},
     {LLM_TENSOR_NEXTN_SHARED_HEAD_NORM,     {LLM_TENSOR_LAYER_REPEATING, GGML_OP_MUL}},
     // Nemotron 3 Super
-    {LLM_TENSOR_FFN_LATENT_DOWN,            {LLM_TENSOR_LAYER_REPEATING, GGML_OP_MUL}},
-    {LLM_TENSOR_FFN_LATENT_UP,              {LLM_TENSOR_LAYER_REPEATING, GGML_OP_MUL}},
+    // latent projections feed ggml_mul_mat, the buft probe must use MUL_MAT to keep them on GPU
+    {LLM_TENSOR_FFN_LATENT_DOWN,            {LLM_TENSOR_LAYER_REPEATING, GGML_OP_MUL_MAT}},
+    {LLM_TENSOR_FFN_LATENT_UP,              {LLM_TENSOR_LAYER_REPEATING, GGML_OP_MUL_MAT}},
 };
 
 LLM_KV::LLM_KV(llm_arch arch, const char * suffix) : arch(arch), suffix(suffix) {}