diff --git a/src/llama.cpp b/src/llama.cpp
index 4f1447d79..9654ffad3 100644
--- a/src/llama.cpp
+++ b/src/llama.cpp
@@ -17402,11 +17402,12 @@ struct llama_context * llama_new_context_with_model(
         params.flash_attn = false;
     }
 
-    if (params.flash_attn && model->arch == LLM_ARCH_GEMMA2) {
-        LLAMA_LOG_WARN("%s: flash_attn is not compatible with Gemma2 - forcing off\n", __func__);
+    if (params.flash_attn && model->hparams.attn_soft_cap) {
+        LLAMA_LOG_WARN("%s: flash_attn is not compatible with attn_soft_cap - forcing off\n", __func__);
         params.flash_attn = false;
     }
 
+
     if (params.flash_attn && model->hparams.n_embd_head_k != model->hparams.n_embd_head_v) {
         LLAMA_LOG_WARN("%s: flash_attn requires n_embd_head_k == n_embd_head_v - forcing off\n", __func__);
         params.flash_attn = false;