diff --git a/inference-ollama-gradio-chat/gradio_chat.py b/inference-ollama-gradio-chat/gradio_chat.py
index 3e4ca79..980cfbd 100644
--- a/inference-ollama-gradio-chat/gradio_chat.py
+++ b/inference-ollama-gradio-chat/gradio_chat.py
@@ -3,25 +3,77 @@
 import os
 
 import gradio as gr
+import requests
 from openai import OpenAI
+from sshtunnel import SSHTunnelForwarder
 
 OLLAMA_BASE_URL = os.environ.get("OLLAMA_BASE_URL", "http://127.0.0.1:11434")
 MODEL_NAME = os.environ.get("MODEL_NAME", "llama2")
 
+# SSH tunneling setup if remote Ollama
+SSH_HOST = os.environ.get("SSH_HOST")
+SSH_USER = os.environ.get("SSH_USER")
+SSH_KEY_PATH = os.environ.get("SSH_KEY_PATH")
+SSH_PORT = int(os.environ.get("SSH_PORT", 22))
+
+tunnel = None
+if SSH_HOST:
+    tunnel = SSHTunnelForwarder(
+        (SSH_HOST, SSH_PORT),
+        ssh_username=SSH_USER,
+        ssh_pkey=SSH_KEY_PATH,
+        remote_bind_address=("127.0.0.1", 11434),
+        local_bind_address=("127.0.0.1", 0),  # 0 means auto-assign port
+    )
+    tunnel.start()
+    local_port = tunnel.local_bind_port
+    OLLAMA_BASE_URL = f"http://127.0.0.1:{local_port}"
+
 client = OpenAI(base_url=f"{OLLAMA_BASE_URL}/v1", api_key="ollama")
 
 
 def chat(message: str, history: list[dict]) -> str:
     messages = history + [{"role": "user", "content": message}]
     response = client.chat.completions.create(model=MODEL_NAME, messages=messages)
-    return response.choices[0].message.content
+    return response.choices[0].message.content or ""
+
+
+def tokenize_text(text: str) -> str:
+    try:
+        response = requests.post(
+            f"{OLLAMA_BASE_URL}/api/tokenize",
+            json={"model": MODEL_NAME, "prompt": text},
+        )
+        if response.status_code == 200:
+            tokens = response.json()["tokens"]
+            return f"Tokens: {tokens}\n\nCount: {len(tokens)}"
+        else:
+            return f"Error: {response.status_code} - {response.text}"
+    except Exception as e:
+        return f"Error: {str(e)}"
 
 
-demo = gr.ChatInterface(
+chat_interface = gr.ChatInterface(
     fn=chat,
     title="Ollama Chat",
     description=f"Chatting with **{MODEL_NAME}** via Ollama",
 )
 
+tokenization_interface = gr.Interface(
+    fn=tokenize_text,
+    inputs=gr.Textbox(label="Input Text", placeholder="Enter text to tokenize"),
+    outputs=gr.Textbox(label="Tokenization Result"),
+    title="Tokenization Preview",
+    description=f"Preview how **{MODEL_NAME}** tokenizes text via Ollama",
+)
+
+demo = gr.TabbedInterface(
+    [chat_interface, tokenization_interface], ["Chat", "Tokenization"]
+)
+
 if __name__ == "__main__":
-    demo.launch(server_name="0.0.0.0", server_port=7860)
+    try:
+        demo.launch(server_name="0.0.0.0", server_port=7860)
+    finally:
+        if tunnel:
+            tunnel.close()
diff --git a/inference-ollama-gradio-chat/task.yaml b/inference-ollama-gradio-chat/task.yaml
index 298d62d..4fd212c 100644
--- a/inference-ollama-gradio-chat/task.yaml
+++ b/inference-ollama-gradio-chat/task.yaml
@@ -5,7 +5,7 @@ resources:
   cpus: 2
   memory: 4
 setup: |
-  uv pip install --upgrade gradio openai
+  uv pip install --upgrade gradio openai sshtunnel requests
   python -c "import gradio; print('gradio installed at:', gradio.__file__)"
   if ! command -v ollama >/dev/null 2>&1; then curl -fsSL https://ollama.com/install.sh | sh; fi
   if [ "$(uname)" != "Darwin" ]; then $SUDO apt-get update && $SUDO apt-get install -y pciutils lshw; fi
@@ -13,9 +13,7 @@ run: |
   export OLLAMA_HOST=0.0.0.0:11434
   ollama serve > /tmp/ollama.log 2>&1 &
   sleep 3
-  ollama pull $MODEL_NAME > /tmp/ollama-pull.log 2>&1 &
-  sleep 5
-  python -c "import gradio; print('gradio found at:', gradio.__file__)" 2>&1 || echo "gradio NOT found in run python"
+  ollama pull $MODEL_NAME
   OLLAMA_BASE_URL=http://127.0.0.1:11434 MODEL_NAME=$MODEL_NAME python ~/inference-ollama-gradio-chat/gradio_chat.py > /tmp/gradio.log 2>&1 &
   sleep 5
   tail -f /tmp/ollama.log /tmp/ollama-pull.log /tmp/gradio.log /tmp/ngrok.log