diff --git a/backends/cuda/cuda_backend.py b/backends/cuda/cuda_backend.py
index 8ed8cdefbb1..ef98de29f23 100644
--- a/backends/cuda/cuda_backend.py
+++ b/backends/cuda/cuda_backend.py
@@ -129,6 +129,8 @@ def preprocess(
                 user_input_placeholders.append(node.meta["val"])
 
         options: dict[str, typing.Any] = {
+            # Better model precision
+            "emulate_precision_casts": True,
             # Embed CUDA kernel binaries directly into the compiled shared object
             "aot_inductor.embed_kernel_binary": True,
             # Do not link against the full PyTorch/libtorch library