LLM Search

​

1️⃣ Real-time Web Search: Breaking LLM Time Limitations for More Accurate and Reliable Outputs

We’ve enhanced OpenAI and Gemini series models with the ability to access the latest information from the web, helping you:

• ✅ Access Latest Information: Get real-time updates on current events, latest research, or live data
• ✅ Eliminate Knowledge Gaps: Overcome the time limitations of LLM training data by accessing post-training information
• ✅ Reduce Hallucinations: Provide fact-based answers through real-time web searches, significantly reducing AI confabulations
• ✅ Improve Decision Quality: Make more confident decisions based on analysis and recommendations grounded in current facts

Supported Models: Currently supporting OpenAI and Gemini model series with two integration methods:

1. Models with Native Search Capabilities Gemini Series (Ground with Google search):

• gemini-2.0-pro-exp-02-05-search
• gemini-2.0-flash-exp-search
• gemini-2.0-flash-search

OpenAI Series (Bing search):

• gpt-4o-search-preview
• gpt-4o-mini-search-preview

2. Parameter-Based Support Simply add the parameter web_search_options={} to enable web connectivity for all Gemini and OpenAI models. The search fee for Gemini models is $3.5 per thousand searches.

​

Usage Guide

Before using, run pip install -U openai to upgrade the openai package.

Example:

from openai import OpenAI

client = OpenAI(
    api_key="AIHUBMIX_API_KEY", # Replace with the key you generated in AiHubMix
    base_url="https://aihubmix.com/v1"
)

chat_completion = client.chat.completions.create(
    model="gemini-2.0-flash-exp",
    # 🌐 Enable search
    web_search_options={},
    messages=[
        {
            "role": "user",
            "content": [
                {
                    "type": "text",
                    "text": "Search for information about the AIhubmix LLM API platform, provide a brief introduction, and include relevant links."
                }
            ]
        }
    ]
)

print(chat_completion.choices[0].message.content)

​

2️⃣ Smart Surfing: Allowing AI to Explore the Internet Freely

By appending :surfing to the model id, any large language model can be equipped with search capabilities.

• Simply append the suffix, no complex integration is required
• This method will default to forwarding the user’s request to the Tavily search service, and the LLM will reference the search results for response
• Search fee: $0.006 per search
• The fee is currently deducted directly from the “credit change”, and the “log detail” does not list the search fee yet, but will be shown in the future

Example:

import requests
import json
import os

try:
    response = requests.post(
        url="https://aihubmix.com/v1/chat/completions",
        headers={
            "Authorization": f"Bearer {os.environ.get('AIHUBMIX_API_KEY')}",
            "Content-Type": "application/json",
        },
        data=json.dumps({
            "model": "gpt-4o-mini:surfing", # Append :surfing to the model id to support searching
            "messages": [
                {
                    "role": "user",
                    "content": "Search the last fact about ChatGPT memory feature, return with the URL"
                }
            ]
        })
    )

    result = response.json()
    print("API response:", json.dumps(result, ensure_ascii=False, indent=2))

except requests.exceptions.RequestException as e:
    print(f"Request error: {e}")
except json.JSONDecodeError as e:
    print(f"JSON decode error: {e}")
except Exception as e:
    print(f"Other error: {e}")

API Response Example:

{
  "id": "chatcmpl-BLN21dGcrv8MrbeHfForjY4bYZHBF",
  "model": "gpt-4o-mini-2024-07-18",
  "object": "chat.completion",
  "created": 1744433121,
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "The latest update about ChatGPT's memory feature indicates that it now has expanded memory capabilities that allow it to recall information in two ways: through “saved memories” that users manually ask it to remember, and “reference chat history” for improving future interactions. For more details, you can visit the URL: [The Verge Article](https://www.theverge.com/news/646968/openai-chatgpt-long-term-memory-upgrade)."
      },
      "finish_reason": "stop"
    }
  ],
  "system_fingerprint": "fp_ded0d14823",
  "usage": {
    "prompt_tokens": 604,
    "completion_tokens": 89,
    "total_tokens": 693,
    "prompt_tokens_details": {
      "audio_tokens": 0,
      "cached_tokens": 0
    },
    "completion_tokens_details": {
      "accepted_prediction_tokens": 0,
      "audio_tokens": 0,
      "reasoning_tokens": 0,
      "rejected_prediction_tokens": 0
    }
  }
}