Serper

Intro

serper is an API service that provides you fast results from Google Search

Features

Response Time: Queries are generally returned within 1 to 2 seconds; occasionally, it can take 2 to 4 seconds if a retry is needed.
Query Rate Limit: The default limit for Ultimate credits is 300 queries per second, allowing 15,000 to 18,000 searches per minute. Higher concurrency limits are available upon request.
Real-Time Queries: All API calls query Google directly for the latest results with no caching involved.
Search Query Customization: Users can specify search locations by country, language, and specific areas such as cities or neighborhoods.
Credit Deduction: Credits are deducted for each successful query response. When the credit balance is zero, new queries will not be accepted.

Example

Create the following python file test.py

import requests
import json

url = "https://google.serper.dev/search"

payload = json.dumps({
  "q": "apple inc"
})
headers = {
  'X-API-KEY': 'API_KEY',
  'Content-Type': 'application/json'
}

response = requests.request("POST", url, headers=headers, data=payload)

print(response.text)

You can clone the following repo which as this example code.

git clone git@github.com:rramos/serper-test.git
cd repo
python -m venv venv
source venv/bin/activate
pip install -r requirements.txt
python test.py

NOTE: Make sure to update the .env value with your API KEY

Playground

On the provided Playground, you will find several examples for multiple languages and various options to apply on your request.

Dashboard

The Dashboard provides you with great visibility on the credits consumption

Langchain Integration

There is a Langchain wrapper if you would like to incorporate the search results.
Here is one example using Ollama. Check the following article if you would like to setup Ollama locally.

import os
from dotenv import load_dotenv

from langchain.agents import AgentType, initialize_agent
from langchain_community.utilities import GoogleSerperAPIWrapper
from langchain_core.tools import Tool
from langchain_community.llms import Ollama

# Load environment variables from .env file
load_dotenv()

# Load API Key
api_key = os.getenv('SERPER_API_KEY')

# Query
query="rramos github tech and data notes"

llm = Ollama(temperature=0,model="llama3")
search = GoogleSerperAPIWrapper()
tools = [
    Tool(
        name="Intermediate Answer",
        func=search.run(query),
        description="useful for when you need to ask with search",
    )
]

self_ask_with_search = initialize_agent(
    tools, llm, agent=AgentType.SELF_ASK_WITH_SEARCH, verbose=True
)
self_ask_with_search.invoke(
    "resume the search results ?"
)

Conclusion

This service is incredible, offering blazing fast results with real-time queries and a user-friendly interface. The dashboard and playground provide all the information you need to get started in a raw simple form which I really like. As for pricing, it’s essential to calculate your costs to determine if integrating this service into your product is cost-effective.

13 Jun, 2024 - About 2 minutes 