diff --git a/unsloth/models/llama.py b/unsloth/models/llama.py
index aedc935d75..f5dc02704e 100644
--- a/unsloth/models/llama.py
+++ b/unsloth/models/llama.py
@@ -65,7 +65,7 @@ from huggingface_hub.utils._token import get_token
 from transformers import __version__ as transformers_version
 from packaging.version import Version
 transformers_version = Version(transformers_version)
-SUPPORTS_LLAMA32 = transformers_version >= Version("4.46")
+SUPPORTS_LLAMA32 = transformers_version > Version("4.45.0")
 
 def original_apply_qkv(self, X):
     Q = self.q_proj(X)
diff --git a/unsloth/models/loader.py b/unsloth/models/loader.py
index 8afb3a3a82..ee1f680ace 100644
--- a/unsloth/models/loader.py
+++ b/unsloth/models/loader.py
@@ -31,7 +31,7 @@ SUPPORTS_FOURBIT = transformers_version >= Version("4.37")
 SUPPORTS_GEMMA   = transformers_version >= Version("4.38")
 SUPPORTS_GEMMA2  = transformers_version >= Version("4.42")
 SUPPORTS_LLAMA31 = transformers_version >= Version("4.43.2")
-SUPPORTS_LLAMA32 = transformers_version >= Version("4.46")
+SUPPORTS_LLAMA32 = transformers_version  > Version("4.45.0")
 if SUPPORTS_GEMMA:
     from .gemma  import FastGemmaModel
 if SUPPORTS_GEMMA2:
@@ -264,7 +264,7 @@ class FastLanguageModel(FastLlamaModel):
                 )
 
             dispatch_model = FastLlamaModel
-            
+
         elif model_type == "mistral": dispatch_model = FastMistralModel
         elif model_type == "gemma":
             if not SUPPORTS_GEMMA: