diff --git a/comfy/diffusers_load.py b/comfy/diffusers_load.py
index 98b888a19..56e63a756 100644
--- a/comfy/diffusers_load.py
+++ b/comfy/diffusers_load.py
@@ -22,7 +22,7 @@ def load_diffusers(model_path, output_vae=True, output_clip=True, embedding_dire
     if text_encoder2_path is not None:
         text_encoder_paths.append(text_encoder2_path)
 
-    unet = comfy.sd.load_unet(unet_path)
+    unet = comfy.sd.load_diffusion_model(unet_path)
 
     clip = None
     if output_clip:
diff --git a/comfy/ldm/flux/layers.py b/comfy/ldm/flux/layers.py
index a7b0c55fa..da0cf61b1 100644
--- a/comfy/ldm/flux/layers.py
+++ b/comfy/ldm/flux/layers.py
@@ -170,8 +170,8 @@ class DoubleStreamBlock(nn.Module):
         txt_attn, img_attn = attn[:, : txt.shape[1]], attn[:, txt.shape[1] :]
 
         # calculate the img bloks
-        img += img_mod1.gate * self.img_attn.proj(img_attn)
-        img += img_mod2.gate * self.img_mlp((1 + img_mod2.scale) * self.img_norm2(img) + img_mod2.shift)
+        img = img + img_mod1.gate * self.img_attn.proj(img_attn)
+        img = img + img_mod2.gate * self.img_mlp((1 + img_mod2.scale) * self.img_norm2(img) + img_mod2.shift)
 
         # calculate the txt bloks
         txt += txt_mod1.gate * self.txt_attn.proj(txt_attn)
diff --git a/comfy/model_management.py b/comfy/model_management.py
index fdf3308ba..152a76f37 100644
--- a/comfy/model_management.py
+++ b/comfy/model_management.py
@@ -434,7 +434,7 @@ def free_memory(memory_required, device, keep_loaded=[]):
                 soft_empty_cache()
     return unloaded_models
 
-def load_models_gpu(models, memory_required=0, force_patch_weights=False, minimum_memory_required=None):
+def load_models_gpu(models, memory_required=0, force_patch_weights=False, minimum_memory_required=None, force_full_load=False):
     global vram_state
 
     inference_memory = minimum_inference_memory()
@@ -513,7 +513,7 @@ def load_models_gpu(models, memory_required=0, force_patch_weights=False, minimu
         else:
             vram_set_state = vram_state
         lowvram_model_memory = 0
-        if lowvram_available and (vram_set_state == VRAMState.LOW_VRAM or vram_set_state == VRAMState.NORMAL_VRAM):
+        if lowvram_available and (vram_set_state == VRAMState.LOW_VRAM or vram_set_state == VRAMState.NORMAL_VRAM) and not force_full_load:
             model_size = loaded_model.model_memory_required(torch_dev)
             current_free_mem = get_free_memory(torch_dev)
             lowvram_model_memory = max(64 * (1024 * 1024), (current_free_mem - minimum_memory_required), min(current_free_mem * 0.4, current_free_mem - minimum_inference_memory()))
diff --git a/comfy/model_patcher.py b/comfy/model_patcher.py
index 39d00d82e..1edbf24ab 100644
--- a/comfy/model_patcher.py
+++ b/comfy/model_patcher.py
@@ -411,7 +411,7 @@ class ModelPatcher:
                     logging.debug("lowvram: loaded module regularly {} {}".format(n, m))
 
         if lowvram_counter > 0:
-            logging.info("loaded partially {} {}".format(lowvram_model_memory / (1024 * 1024), patch_counter))
+            logging.info("loaded partially {} {} {}".format(lowvram_model_memory / (1024 * 1024), mem_counter / (1024 * 1024), patch_counter))
             self.model.model_lowvram = True
         else:
             logging.info("loaded completely {} {}".format(lowvram_model_memory / (1024 * 1024), mem_counter / (1024 * 1024)))
diff --git a/comfy/sd.py b/comfy/sd.py
index c8a2f086c..edd0b51d8 100644
--- a/comfy/sd.py
+++ b/comfy/sd.py
@@ -86,7 +86,7 @@ class CLIP:
         self.tokenizer = tokenizer(embedding_directory=embedding_directory, tokenizer_data=tokenizer_data)
         self.patcher = comfy.model_patcher.ModelPatcher(self.cond_stage_model, load_device=load_device, offload_device=offload_device)
         if params['device'] == load_device:
-            model_management.load_model_gpu(self.patcher)
+            model_management.load_models_gpu([self.patcher], force_full_load=True)
         self.layer_idx = None
         logging.debug("CLIP model load device: {}, offload device: {}, current: {}".format(load_device, offload_device, params['device']))
 
@@ -585,12 +585,13 @@ def load_state_dict_guess_config(sd, output_vae=True, output_clip=True, output_c
         model_patcher = comfy.model_patcher.ModelPatcher(model, load_device=load_device, offload_device=model_management.unet_offload_device())
         if inital_load_device != torch.device("cpu"):
             logging.info("loaded straight to GPU")
-            model_management.load_model_gpu(model_patcher)
+            model_management.load_models_gpu([model_patcher], force_full_load=True)
 
     return (model_patcher, clip, vae, clipvision)
 
 
-def load_unet_state_dict(sd, dtype=None): #load unet in diffusers or regular format
+def load_diffusion_model_state_dict(sd, model_options={}): #load unet in diffusers or regular format
+    dtype = model_options.get("dtype", None)
 
     #Allow loading unets from checkpoint files
     diffusion_model_prefix = model_detection.unet_prefix_from_state_dict(sd)
@@ -632,6 +633,7 @@ def load_unet_state_dict(sd, dtype=None): #load unet in diffusers or regular for
 
     manual_cast_dtype = model_management.unet_manual_cast(unet_dtype, load_device, model_config.supported_inference_dtypes)
     model_config.set_inference_dtype(unet_dtype, manual_cast_dtype)
+    model_config.custom_operations = model_options.get("custom_operations", None)
     model = model_config.get_model(new_sd, "")
     model = model.to(offload_device)
     model.load_model_weights(new_sd, "")
@@ -640,14 +642,23 @@ def load_unet_state_dict(sd, dtype=None): #load unet in diffusers or regular for
         logging.info("left over keys in unet: {}".format(left_over))
     return comfy.model_patcher.ModelPatcher(model, load_device=load_device, offload_device=offload_device)
 
-def load_unet(unet_path, dtype=None):
+
+def load_diffusion_model(unet_path, model_options={}):
     sd = comfy.utils.load_torch_file(unet_path)
-    model = load_unet_state_dict(sd, dtype=dtype)
+    model = load_diffusion_model_state_dict(sd, model_options=model_options)
     if model is None:
         logging.error("ERROR UNSUPPORTED UNET {}".format(unet_path))
         raise RuntimeError("ERROR: Could not detect model type of: {}".format(unet_path))
     return model
 
+def load_unet(unet_path, dtype=None):
+    print("WARNING: the load_unet function has been deprecated and will be removed please switch to: load_diffusion_model")
+    return load_diffusion_model(unet_path, model_options={"dtype": dtype})
+
+def load_unet_state_dict(sd, dtype=None):
+    print("WARNING: the load_unet_state_dict function has been deprecated and will be removed please switch to: load_diffusion_model_state_dict")
+    return load_diffusion_model_state_dict(sd, model_options={"dtype": dtype})
+
 def save_checkpoint(output_path, model, clip=None, vae=None, clip_vision=None, metadata=None, extra_keys={}):
     clip_sd = None
     load_models = [model]
diff --git a/nodes.py b/nodes.py
index e296597c5..525b28d85 100644
--- a/nodes.py
+++ b/nodes.py
@@ -826,14 +826,14 @@ class UNETLoader:
     CATEGORY = "advanced/loaders"
 
     def load_unet(self, unet_name, weight_dtype):
-        dtype = None
+        model_options = {}
         if weight_dtype == "fp8_e4m3fn":
-            dtype = torch.float8_e4m3fn
+            model_options["dtype"] = torch.float8_e4m3fn
         elif weight_dtype == "fp8_e5m2":
-            dtype = torch.float8_e5m2
+            model_options["dtype"] = torch.float8_e5m2
 
         unet_path = folder_paths.get_full_path("unet", unet_name)
-        model = comfy.sd.load_unet(unet_path, dtype=dtype)
+        model = comfy.sd.load_diffusion_model(unet_path, model_options=model_options)
         return (model,)
 
 class CLIPLoader: