LlamaIndex

1. Preparation

• Log in to cometapi. Click "ADD API key" in the API keys to get your token key: sk-xxxxx

Prerequisites

Python 3.8+ installed.

A CometAPI account. Generate an API key from the API Keys page.

Jupyter Notebook or Python environment (Google Colab recommended for interactive testing).

Step 1: Install LlamaIndex

Install the LlamaIndex library and CometAPI integration using pip. This is a one-time setup.

Step 2: Set Up API Key

You need the key obtained from CometAPI to authenticate requests. You can set it in environment variables or pass it in the class constructor.

Python example:

Note: Using environment variables is more secure to avoid hardcoding sensitive information in scripts.

Step 3: Perform Basic Completion Calls

Initialize the model using LlamaIndex's CometLLM class and call the chat or complete methods. You can specify models like gpt-5-chat-latest or gpt-4o.

Example:

Expected Output:

Chat response: e.g., assistant: Hi

Completion response: e.g., a text description about Kaiming He, including information on ResNet.

This sends a simple message and retrieves the model response. You can customize messages for more complex interactions.

Step 4: Streaming Calls

For real-time applications, use the stream_chat or stream_complete methods for streaming responses.

Example:

Explanation:

stream_chat and stream_complete generate responses chunk by chunk, suitable for real-time output.

If an error occurs, it will be displayed in the console.

Expected Output:

Streaming printed response content, e.g., an explanation of ResNet or an overview of large language models, appearing in chunks.

Additional Tips

Supported Models: CometAPI supports various models such as gpt-5-chat-latest, gpt-4o, claude-3-7-sonnet-latest, etc. Check the CometAPI documentation for the latest list.

Error Handling: Always wrap calls in try-except blocks to handle invalid keys or network errors.

Advanced Features: LlamaIndex supports parameters like max_tokens and temperature for fine-tuning responses. Add them during CometLLM initialization.

Security: Never commit API keys to version control. Use environment variables or secret managers.

Troubleshooting: If issues arise, ensure the API key is valid and check LlamaIndex logs. For more details, refer to the LlamaIndex documentation or CometAPI documentation or Colab Example.

Other Models: For example, using a Claude model: claude_llm = CometLLM(api_key=api_key, model="claude-3-7-sonnet-latest", max_tokens=1024).

Rate Limits and Costs: Monitor usage in the CometAPI console.

Code Example

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "initial-install",
   "metadata": {},
   "outputs": [],
   "source": [
    "%pip install llama-index-llms-cometapi\n",
    "%pip install llama-index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "setup-api-key",
   "metadata": {},
   "outputs": [],
   "source": [
    "from llama_index.llms.cometapi import CometLLM\n",
    "import os\n",
    "\n",
    "# Set API key\n",
    "\n",
    "# Open the [API Keys](https://api.cometapi.com/console/token) page in CometAPI.\n",
    "# Create a new key or copy an existing one,\n",
    "# then paste it here.\n",
    "os.environ[\"COMETAPI_KEY\"] = ''\n",
    "api_key = os.getenv(\"COMETAPI_KEY\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "basic-calls",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Initialize LLM\n",
    "llm = CometLLM(\n",
    "    api_key=api_key,\n",
    "    max_tokens=256,\n",
    "    context_window=4096,\n",
    "    model=\"gpt-5-chat-latest\",\n",
    ")\n",
    "\n",
    "# Chat call using ChatMessage\n",
    "from llama_index.core.llms import ChatMessage\n",
    "\n",
    "messages = [\n",
    "    ChatMessage(role=\"system\", content=\"You are a helpful assistant\"),\n",
    "    ChatMessage(role=\"user\", content=\"Say 'Hi' only!\"),\n",
    "]\n",
    "resp = llm.chat(messages)\n",
    "print(resp)\n",
    "\n",
    "# Use complete method\n",
    "resp = llm.complete(\"Who is Kaiming He\")\n",
    "print(resp)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "streaming-calls",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Streaming chat\n",
    "message = ChatMessage(role=\"user\", content=\"Tell me what ResNet is\")\n",
    "resp = llm.stream_chat([message])\n",
    "for r in resp:\n",
    "    print(r.delta, end=\"\")\n",
    "\n",
    "# Streaming completion\n",
    "resp = llm.stream_complete(\"Tell me about Large Language Models\")\n",
    "for r in resp:\n",
    "    print(r.delta, end=\"\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "different-model",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Use Claude model\n",
    "claude_llm = CometLLM(\n",
    "    api_key=api_key, model=\"claude-3-7-sonnet-latest\", max_tokens=200\n",
    ")\n",
    "\n",
    "resp = claude_llm.complete(\"Explain deep learning briefly\")\n",
    "print(resp)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8+"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}

1. Preparation#

Prerequisites#

Step 1: Install LlamaIndex#

Step 2: Set Up API Key#

Step 3: Perform Basic Completion Calls#

Step 4: Streaming Calls#