diff --git a/unsloth/models/_utils.py b/unsloth/models/_utils.py
index 63f48af659..747858d011 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__ = "2025.5.5"
+__version__ = "2025.5.6"
 
 __all__ = [
     "SUPPORTS_BFLOAT16",
diff --git a/unsloth/models/loader.py b/unsloth/models/loader.py
index 9c5a7b68be..8a49026984 100644
--- a/unsloth/models/loader.py
+++ b/unsloth/models/loader.py
@@ -461,6 +461,12 @@ except:
     from transformers import AutoModelForVision2Seq
 pass
 
+DISABLE_COMPILE_MODEL_NAMES = [
+    "aya-vision",
+    "modernbert",
+    "granite-vision",
+]
+
 
 class FastModel(FastBaseModel):
     @staticmethod
@@ -521,33 +527,35 @@ class FastModel(FastBaseModel):
             model_name = get_model_name(model_name, load_in_4bit)
 
         # Check versions
+        lowered_model_name = model_name.lower()
         LATEST  = '\nPlease use transformers via `pip install --no-deps git+https://github.com/huggingface/transformers.git`'
         NIGHTLY = '\nPlease use nightly transformers via pip install --upgrade "transformers>=4.49.0"`'
-        if "pixtral" in model_name.lower() and transformers_version < Version("4.49.0"):
+        if "pixtral" in lowered_model_name and transformers_version < Version("4.49.0"):
             raise RuntimeError("Unsloth: Pixtral only works on transformers >= 4.49.0." + LATEST)
-        elif "qwen2.5" in model_name.lower() and transformers_version < Version("4.49.0"):
+        elif "qwen2.5" in lowered_model_name and transformers_version < Version("4.49.0"):
             raise RuntimeError("Unsloth: Qwen 2.5 only works on transformers >= 4.49.0." + LATEST)
-        elif "aya-vision" in model_name.lower():
-            # Disable compiling for now - errors out!
-            os.environ["UNSLOTH_COMPILE_DISABLE"] = "1"
-            if transformers_version < Version("4.50.0.dev0"):
-                raise RuntimeError("Unsloth: Aya Vision only works on transformers >= 4.50.0." + NIGHTLY)
-        elif "gemma-3" in model_name.lower() and transformers_version < Version("4.50.0.dev0"):
+        elif "gemma-3" in lowered_model_name and transformers_version < Version("4.50.0.dev0"):
             raise RuntimeError("Unsloth: Gemma 3 only works on transformers >= 4.50.0." + NIGHTLY)
-        elif "c4ai-command-a-03-2025" in model_name.lower() and transformers_version < Version("4.50.0.dev0"):
+        elif "c4ai-command-a-03-2025" in lowered_model_name and transformers_version < Version("4.50.0.dev0"):
             raise RuntimeError("Unsloth: Cohere's Command model only works on transformers >= 4.50.0." + NIGHTLY)
-        elif "granite-vision" in model_name.lower():
-            # Disable compiling for now - errors out!
-            os.environ["UNSLOTH_COMPILE_DISABLE"] = "1"
-            if transformers_version < Version("4.50.0.dev0"):
-                raise RuntimeError("Unsloth: Granite Vision only works on transformers >= 4.50.0." + NIGHTLY)
-        elif "csm-1b" in model_name.lower():
+        elif "csm-1b" in lowered_model_name:
             os.environ["UNSLOTH_DISABLE_STATIC_GENERATION"] = "1" # Sesame fails
             os.environ["UNSLOTH_FORCE_CUSTOM_DTYPE"] = "torch.float16;if name.endswith(('_proj', 'fc1', 'fc2', 'codebook', 'head')): module.to(torch.float16)"
-        elif "olmo-2" in model_name.lower() and transformers_version < Version("4.50.0.dev0"):
+        elif "olmo-2" in lowered_model_name and transformers_version < Version("4.50.0.dev0"):
             raise RuntimeError("Unsloth: OLMo-2 only works on transformers >= 4.50.0." + NIGHTLY)
-        elif "whisper" in model_name.lower():
-            os.environ["UNSLOTH_DISABLE_STATIC_GENERATION"] = "1" # Whisper fails
+        else:
+            for check_model_name in DISABLE_COMPILE_MODEL_NAMES:
+                if check_model_name in lowered_model_name:
+                    os.environ["UNSLOTH_COMPILE_DISABLE"] = "1"
+                    os.environ["UNSLOTH_DISABLE_STATIC_GENERATION"] = "1"
+                    if transformers_version < Version("4.50.0.dev0"):
+                        raise RuntimeError(f"Unsloth: {check_model_name} only works on transformers >= 4.50.0." + NIGHTLY)
+                    break
+        pass
+
+        if auto_model is not None:
+            # All other models need to disable static cache
+            os.environ["UNSLOTH_DISABLE_STATIC_GENERATION"] = "1"
         pass
 
         if USE_MODELSCOPE and not os.path.exists(model_name):
diff --git a/unsloth/models/mapper.py b/unsloth/models/mapper.py
index 4bbd8295cc..e50a5a877d 100644
--- a/unsloth/models/mapper.py
+++ b/unsloth/models/mapper.py
@@ -728,16 +728,6 @@ __INT_TO_FLOAT_MAPPER = \
         "mistralai/Mistral-Small-3.1-24B-Base-2503",
         "unsloth/Mistral-Small-3.1-24B-Base-2503-bnb-4bit",
     ),
-    "unsloth/orpheus-3b-0.1-pretrained-unsloth-bnb-4bit" : (
-        "unsloth/orpheus-3b-0.1-pretrained",
-        "canopylabs/orpheus-3b-0.1-pretrained",
-        "unsloth/orpheus-3b-0.1-pretrained-bnb-4bit",
-    ),
-    "unsloth/orpheus-3b-0.1-ft-unsloth-bnb-4bit" : (
-        "unsloth/orpheus-3b-0.1-ft",
-        "canopylabs/orpheus-3b-0.1-ft",
-        "unsloth/orpheus-3b-0.1-ft-bnb-4bit",
-    ),
     "unsloth/Qwen3-0.6B-unsloth-bnb-4bit" : (
         "unsloth/Qwen3-0.6B",
         "Qwen/Qwen3-0.6B",
@@ -817,6 +807,16 @@ __INT_TO_FLOAT_MAPPER = \
         "microsoft/Phi-4-mini-reasoning",
         "unsloth/phi-4-mini-reasoning-bnb-4bit",
     ),
+    "unsloth/orpheus-3b-0.1-pretrained-unsloth-bnb-4bit" : (
+        "unsloth/orpheus-3b-0.1-pretrained",
+        "canopylabs/orpheus-3b-0.1-pretrained",
+        "unsloth/orpheus-3b-0.1-pretrained-bnb-4bit",
+    ),
+    "unsloth/orpheus-3b-0.1-ft-unsloth-bnb-4bit" : (
+        "unsloth/orpheus-3b-0.1-ft",
+        "canopylabs/orpheus-3b-0.1-ft",
+        "unsloth/orpheus-3b-0.1-ft-bnb-4bit",
+    ),
     "unsloth/csm-1b" : (
         "unsloth/csm-1b",
         "sesame/csm-1b",
@@ -837,6 +837,18 @@ __INT_TO_FLOAT_MAPPER = \
         "unsloth/CrisperWhisper",
         "nyrahealth/CrisperWhisper",
     ),
+    "unsloth/Llasa-1B" : (
+        "unsloth/Llasa-1B",
+        "HKUSTAudio/Llasa-1B",
+    ),
+    "unsloth/Spark-TTS-0.5B" : (
+        "unsloth/Spark-TTS-0.5B",
+        "SparkAudio/Spark-TTS-0.5B",
+    ),
+    "unsloth/Llama-OuteTTS-1.0-1B" : (
+        "unsloth/Llama-OuteTTS-1.0-1B",
+        "OuteAI/Llama-OuteTTS-1.0-1B",
+    ),
 }
 
 INT_TO_FLOAT_MAPPER  = {}
diff --git a/unsloth/models/vision.py b/unsloth/models/vision.py
index 2bff87d8d9..4466128a28 100644
--- a/unsloth/models/vision.py
+++ b/unsloth/models/vision.py
@@ -95,8 +95,18 @@ def unsloth_base_fast_generate(
         input_ids = kwargs["input_ids"]
     elif "input" in kwargs:
         input_ids = kwargs["input_ids"]
+    elif "input_features" in kwargs:
+        input_ids = kwargs["input_features"]
+    elif "input_embeds" in kwargs:
+        input_ids = kwargs["input_embeds"]
+    elif "inputs" in kwargs:
+        input_ids = kwargs["inputs"]
     else:
-        raise TypeError("Unsloth: You need to pass in input_ids to .generate!")
+        key = next(iter(kwargs.keys()))
+        if type(kwargs["key"]) is not torch.Tensor:
+            raise TypeError("Unsloth: You need to pass in input_ids to .generate!")
+        input_ids = kwargs[key]
+    pass
     assert(type(input_ids) is torch.Tensor)
     bsz = input_ids.shape[0]
 
@@ -203,10 +213,11 @@ def unsloth_base_fast_generate(
 
     if "generation_config" in kwargs:
         kwargs["generation_config"].cache_implementation = cache_implementation
-        kwargs["generation_config"].compile_config = _compile_config if cache_implementation is not None else None
+        if cache_implementation is not None:
+            kwargs["generation_config"].compile_config = _compile_config
     else:
         kwargs["cache_implementation"] = cache_implementation
-        if cache_implementation:
+        if cache_implementation is not None:
             kwargs["compile_config"] = _compile_config
     pass