diff --git a/pyproject.toml b/pyproject.toml
index 3d381fa669..b3ee2d7aea 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -37,7 +37,7 @@ triton = [
 ]
 
 huggingface = [
-    "unsloth_zoo>=2025.7.7",
+    "unsloth_zoo>=2025.7.8",
     "packaging",
     "tyro",
     "transformers>=4.51.3,!=4.47.0,!=4.52.0,!=4.52.1,!=4.52.2,!=4.52.3,!=4.53.0",
@@ -381,7 +381,7 @@ colab-ampere-torch220 = [
     "flash-attn>=2.6.3",
 ]
 colab-new = [
-    "unsloth_zoo>=2025.7.7",
+    "unsloth_zoo>=2025.7.8",
     "packaging",
     "tyro",
     "transformers>=4.51.3,!=4.47.0,!=4.52.0,!=4.52.1,!=4.52.2,!=4.52.3,!=4.53.0",
diff --git a/unsloth/dataprep/synthetic.py b/unsloth/dataprep/synthetic.py
index f3fbf77222..23268807b1 100644
--- a/unsloth/dataprep/synthetic.py
+++ b/unsloth/dataprep/synthetic.py
@@ -80,6 +80,10 @@ class SyntheticDataKit:
         )
         if "dtype" in engine_args:
             dtype_val = engine_args["dtype"]
+            if   dtype_val == torch.float16:  dtype_val = "float16"
+            elif dtype_val == torch.bfloat16: dtype_val = "bfloat16"
+            elif dtype_val == torch.float32:  dtype_val = "float32"
+            engine_args["dtype"] = dtype_val
             # Convert torch.bfloat16, torch.float16, etc. to valid CLI string
             if hasattr(dtype_val, "name"):
                 engine_args["dtype"] = dtype_val.name
diff --git a/unsloth/models/_utils.py b/unsloth/models/_utils.py
index 1774ed373a..21ebbfd7f4 100644
--- a/unsloth/models/_utils.py
+++ b/unsloth/models/_utils.py
@@ -12,7 +12,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-__version__ = "2025.7.5"
+__version__ = "2025.7.6"
 
 __all__ = [
     "SUPPORTS_BFLOAT16",
diff --git a/unsloth/save.py b/unsloth/save.py
index e61026318a..38c73c9956 100644
--- a/unsloth/save.py
+++ b/unsloth/save.py
@@ -2240,6 +2240,7 @@ from unsloth_zoo.llama_cpp import (
 def save_to_gguf_generic(
     model,
     save_directory,
+    quantization_method = None,
     quantization_type = "Q8_0",
     repo_id = None,
     token = None,
@@ -2252,29 +2253,63 @@ def save_to_gguf_generic(
         install_llama_cpp(just_clone_repo = True)
     pass
 
-    metadata = _convert_to_gguf(
-        save_directory,
-        print_output = True,
-        quantization_type = quantization_type,
-    )
-    if repo_id is not None:
-        prepare_saving(
-            model,
-            repo_id,
-            push_to_hub = True,
-            max_shard_size = "50GB",
-            private = True,
-            token = token,
-        )
+    # Use old style quantization_method
+    new_quantization_methods = []
+    if quantization_method is not None:
+        # Convert quantization_method to list
+        if   isinstance(quantization_method, list):  pass
+        elif isinstance(quantization_method, str):   quantization_method = [ quantization_method, ]
+        elif isinstance(quantization_method, tuple): quantization_method = list(quantization_method)
+        else:
+            raise TypeError("Unsloth: quantization_method can only be a string or a list of strings")
+        pass
+        for i, quant_method in enumerate(quantization_method):
+            quant_method = quant_method.lower()
+            if   quant_method == "not_quantized":  quant_method = "f16"
+            elif quant_method == "fast_quantized": quant_method = "q8_0"
+            elif quant_method == "quantized":      quant_method = "q4_k_m"
+            elif quant_method is None:             quant_method = "q8_0"
+            new_quantization_methods.append(quant_method.lower())
+        pass
+    else:
+        new_quantization_methods.append(quantization_type.lower())
+    # Check if wrong method
+    for quant_method in new_quantization_methods:
+        if quant_method not in ALLOWED_QUANTS.keys():
+            error = f"Unsloth: Quant method = [{quant_method}] not supported. Choose from below:\n"
+            for key, value in ALLOWED_QUANTS.items():
+                error += f"[{key}] => {value}\n"
+            raise RuntimeError(error)
+        pass
+    pass
 
-        from huggingface_hub import HfApi
-        api = HfApi(token = token)
-        api.upload_folder(
-            folder_path = save_directory,
-            repo_id = repo_id,
-            repo_type = "model",
-            allow_patterns = ["*.gguf"],
+    # Go through all types and save individually - somewhat inefficient
+    # since we save F16 / BF16 multiple times
+    for quantization_type in new_quantization_methods:
+        metadata = _convert_to_gguf(
+            save_directory,
+            print_output = True,
+            quantization_type = quantization_type,
         )
+        if repo_id is not None:
+            prepare_saving(
+                model,
+                repo_id,
+                push_to_hub = True,
+                max_shard_size = "50GB",
+                private = True,
+                token = token,
+            )
+
+            from huggingface_hub import HfApi
+            api = HfApi(token = token)
+            api.upload_folder(
+                folder_path = save_directory,
+                repo_id = repo_id,
+                repo_type = "model",
+                allow_patterns = ["*.gguf"],
+            )
+        pass
     pass
     return metadata
 pass