diff --git a/ggml-cuda/fattn-vec-f32.cuh b/ggml-cuda/fattn-vec-f32.cuh
index ce23a4ebd00..ddf0c83740f 100644
--- a/ggml-cuda/fattn-vec-f32.cuh
+++ b/ggml-cuda/fattn-vec-f32.cuh
@@ -278,14 +278,10 @@ void ggml_cuda_flash_attn_ext_vec_f32_case_impl(ggml_backend_cuda_context & ctx,
 
 template <int D, ggml_type type_K, ggml_type type_V>
 void ggml_cuda_flash_attn_ext_vec_f32_case(ggml_backend_cuda_context & ctx, ggml_tensor * dst) {
-    ggml_tensor * KQV = dst;
     ggml_tensor * Q   = dst->src[0];
     ggml_tensor * K   = dst->src[1];
     ggml_tensor * V   = dst->src[2];
 
-    const int32_t precision = KQV->op_params[2];
-    GGML_ASSERT(precision == GGML_PREC_DEFAULT);
-
     GGML_ASSERT(K->type == type_K);
     GGML_ASSERT(V->type == type_V);