diff --git a/web_demo_2.6.py b/web_demo_2.6.py
index 2bebc19..ea58e9a 100644
--- a/web_demo_2.6.py
+++ b/web_demo_2.6.py
@@ -33,6 +33,10 @@
 device = args.device
 assert device in ['cuda', 'mps']
 
+
+TORCH_TYPE = torch.bfloat16 if torch.cuda.is_available() and torch.cuda.get_device_capability()[
+    0] >= 8 else torch.float16
+
 # Load model
 model_path = 'openbmb/MiniCPM-V-2_6'
 if 'int4' in model_path:
@@ -44,7 +48,7 @@
     if args.multi_gpus:
         from accelerate import load_checkpoint_and_dispatch, init_empty_weights, infer_auto_device_map
         with init_empty_weights():
-            model = AutoModel.from_pretrained(model_path, trust_remote_code=True, attn_implementation='sdpa', torch_dtype=torch.bfloat16)
+            model = AutoModel.from_pretrained(model_path, trust_remote_code=True, attn_implementation='sdpa', torch_dtype=TORCH_TYPE)
         device_map = infer_auto_device_map(model, max_memory={0: "10GB", 1: "10GB"},
             no_split_module_classes=['SiglipVisionTransformer', 'Qwen2DecoderLayer'])
         device_id = device_map["llm.model.embed_tokens"]
@@ -63,9 +67,9 @@
         device_map["llm.model.layers.16"] = device_id2
         #print(device_map)
 
-        model = load_checkpoint_and_dispatch(model, model_path, dtype=torch.bfloat16, device_map=device_map)
+        model = load_checkpoint_and_dispatch(model, model_path, dtype=TORCH_TYPE, device_map=device_map)
     else:
-        model = AutoModel.from_pretrained(model_path, trust_remote_code=True, torch_dtype=torch.bfloat16)
+        model = AutoModel.from_pretrained(model_path, trust_remote_code=True, torch_dtype=TORCH_TYPE)
         model = model.to(device=device)
 tokenizer = AutoTokenizer.from_pretrained(model_path, trust_remote_code=True)
 model.eval()
@@ -554,4 +558,3 @@ def select_chat_type(_tab, _app_cfg):
 
 # launch
 demo.launch(share=False, debug=True, show_api=False, server_port=8885, server_name="0.0.0.0")
-