fix: set LLM_NIM_URL for Option 1 to use NVIDIA API endpoint

- When Option 1 is selected (NVIDIA API key for cloud endpoints),
  set LLM_NIM_URL to https://integrate.api.nvidia.com/v1
- This endpoint works with NVIDIA API keys, unlike api.brev.dev
  which requires Brev API keys
- Prevents 401 authentication errors when using NVIDIA API key
  with cloud endpoints
- Update summary output to show LLM_NIM_URL is configured

Fixes: 401 Unauthorized errors when using Option 1 with cloud endpoints

Author: T-DevH

notebooks/setup/complete_setup_guide.ipynb

@@ -1469,12 +1469,15 @@
     "            # Remove EMBEDDING_API_KEY line if using Option 1 (will use NVIDIA_API_KEY)\n",
     "            content = re.sub(r'^EMBEDDING_API_KEY=.*$\\n?', '', content, flags=re.MULTILINE)\n",
     "        \n",
-    "        # Update LLM_MODEL\n",
-    "        # Option 1: Use default cloud model (nvidia/llama-3.3-nemotron-super-49b-v1)\n",
-    "        # Option 2: Use Brev model name if provided\n",
+    "        # Update LLM_MODEL and LLM_NIM_URL\n",
+    "        # Option 1: Use default cloud model with NVIDIA API endpoint (integrate.api.nvidia.com)\n",
+    "        # Option 2: Use Brev model name with Brev endpoint (api.brev.dev)\n",
     "        if choice == \"1\":\n",
-    "            # Option 1: Set to default cloud model\n",
+    "            # Option 1: Set to default cloud model with NVIDIA API endpoint\n",
     "            default_cloud_model = \"nvidia/llama-3.3-nemotron-super-49b-v1\"\n",
+    "            nvidia_llm_endpoint = \"https://integrate.api.nvidia.com/v1\"\n",
+    "            \n",
+    "            # Update LLM_MODEL\n",
     "            if re.search(r'^LLM_MODEL=.*$', content, flags=re.MULTILINE):\n",
     "                content = re.sub(\n",
     "                    r'^LLM_MODEL=.*$',\n",
@@ -1490,6 +1493,23 @@
     "                    content,\n",
     "                    flags=re.MULTILINE\n",
     "                )\n",
+    "            \n",
+    "            # Update LLM_NIM_URL to use NVIDIA API endpoint (works with NVIDIA API keys)\n",
+    "            if re.search(r'^LLM_NIM_URL=.*$', content, flags=re.MULTILINE):\n",
+    "                content = re.sub(\n",
+    "                    r'^LLM_NIM_URL=.*$',\n",
+    "                    f'LLM_NIM_URL={nvidia_llm_endpoint}',\n",
+    "                    content,\n",
+    "                    flags=re.MULTILINE\n",
+    "                )\n",
+    "            else:\n",
+    "                # Add LLM_NIM_URL after LLM_MODEL if it doesn't exist\n",
+    "                content = re.sub(\n",
+    "                    r'^(LLM_MODEL=.*)$',\n",
+    "                    rf'\\1\\nLLM_NIM_URL={nvidia_llm_endpoint}',\n",
+    "                    content,\n",
+    "                    flags=re.MULTILINE\n",
+    "                )\n",
     "        elif brev_model:\n",
     "            # Option 2: Use Brev model name if provided\n",
     "            if re.search(r'^LLM_MODEL=.*$', content, flags=re.MULTILINE):\n",
@@ -1593,6 +1613,7 @@
     "        if choice == \"1\":\n",
     "            print(\"   • NVIDIA_API_KEY: Set (will be used for all services)\")\n",
     "            print(\"   • LLM_MODEL: Set (nvidia/llama-3.3-nemotron-super-49b-v1)\")\n",
+    "            print(\"   • LLM_NIM_URL: Set (https://integrate.api.nvidia.com/v1)\")\n",
     "        else:\n",
     "            print(\"   • NVIDIA_API_KEY: Set (Brev API key for LLM)\")\n",
     "            print(\"   • EMBEDDING_API_KEY: Set (NVIDIA API key for Embedding)\")\n",