Create app.py

app.py

@@ -0,0 +1,35 @@
+import os
+import gradio as gr
+from huggingface_hub import InferenceClient
+
+MODEL = "huihui-ai/Huihui-Qwen3.6-35B-A3B-Claude-4.7-Opus-abliterated"
+
+client = InferenceClient(
+    api_key=os.environ["HF_TOKEN"]
+)
+
+def chat(message, history):
+    messages = []
+
+    for user, assistant in history:
+        messages.append({"role": "user", "content": user})
+        messages.append({"role": "assistant", "content": assistant})
+
+    messages.append({"role": "user", "content": message})
+
+    response = ""
+
+    for chunk in client.chat.completions.create(
+        model=MODEL,
+        messages=messages,
+        stream=True,
+    ):
+        delta = chunk.choices[0].delta.content
+        if delta:
+            response += delta
+            yield response
+
+gr.ChatInterface(
+    fn=chat,
+    title="Huihui-Qwen3.6-35B-A3B-Claude-4.7-Opus-abliterated",
+).launch()