diff --git a/extensions-builtin/sdnext-modernui b/extensions-builtin/sdnext-modernui
index 2463e18fb..d92c929c9 160000
--- a/extensions-builtin/sdnext-modernui
+++ b/extensions-builtin/sdnext-modernui
@@ -1 +1 @@
-Subproject commit 2463e18fb54e1e6aa63fa2233e676e8dd0bf9086
+Subproject commit d92c929c929f8e745bf2f37e4e92aa558bf64de6
diff --git a/modules/minimax/minimax_ui.py b/modules/minimax/minimax_ui.py
index 1f87d3412..21378e4b5 100644
--- a/modules/minimax/minimax_ui.py
+++ b/modules/minimax/minimax_ui.py
@@ -20,7 +20,7 @@ def create_ui(prompt, _negative, styles, overrides, script_inputs, mp4_fps, mp4_
                 model = gr.Dropdown(label='MiniMax model', choices=minimax_models, value=minimax_models[0], elem_id="minimax_model")
                 btn_load = ToolButton(ui_symbols.loading, elem_id="video_model_load_minimax")
             with gr.Row():
-                workflow = gr.Label(value='', label='Workflow', elem_id='minimax_workflow', show_label=False)
+                workflow = gr.Label(value='', label='Workflow', elem_id='minimax_workflow', show_label=False, elem_classes=['highlighted-label'])
             with gr.Accordion(open=True, label='Parameters', elem_id='minimax_param_accordion') as _param_accordion:
                 with gr.Row():
                     width, height = ui_sections.create_resolution_inputs('minimax', default_width=832, default_height=480, step=32)
diff --git a/modules/minimax/minimax_video.py b/modules/minimax/minimax_video.py
index 00d76ccce..aa1a41136 100644
--- a/modules/minimax/minimax_video.py
+++ b/modules/minimax/minimax_video.py
@@ -37,7 +37,7 @@ def load_model(model: str):
     return None
 
 
-def prepare_inputs(workflow: str, p: processing.StableDiffusionProcessingVideo, init_image, last_image, reference_media):
+def prepare_inputs(workflow: str, p: processing.StableDiffusionProcessingVideo, init_image: Image.Image | None, last_image: Image.Image | None, reference_media: list | None):
     from diffusers.modular_pipelines.minimax_h3 import MiniMaxH3ImageReference, MiniMaxH3VideoReference, MiniMaxH3AudioReference
     if workflow == 'fl2va':
         if init_image is not None:
@@ -46,6 +46,8 @@ def prepare_inputs(workflow: str, p: processing.StableDiffusionProcessingVideo,
             p.task_args['last_image'] = last_image
         log.debug(f'Prepare inputs: workflow={workflow} first={init_image} last={last_image}')
     if workflow == 'ref2va':
+        if reference_media is None or len(reference_media) == 0:
+            return
         files = []
         references = []
         for fn in reference_media:
@@ -72,10 +74,13 @@ def prepare_inputs(workflow: str, p: processing.StableDiffusionProcessingVideo,
 
 
 def generate(task_id, _ui_state,
-             model, workflow,
+             model,
+             workflow,
              prompt, styles,
-             width, height, frames,
-             steps, seed,
+             width, height,
+             frames,
+             steps,
+             seed,
              init_image, last_image, reference_media,
              mp4_fps, mp4_interpolate, mp4_codec, mp4_ext, mp4_opt, mp4_video, mp4_frames, mp4_sf, mp4_thumb,
              audio_enable,
@@ -125,6 +130,8 @@ def generate(task_id, _ui_state,
         _processed: processing.Processed = scripts_manager.scripts_video.run(p, *args)
         processed = processing.process_images(p)
 
+        sd_models.offload_ondemand(shared.sd_model, reason='finish')
+
         # init vars
         pixels = None
         num_frames = 0