Baidu Qianfan API - Permanently Free Chinese AI API Service

📋 Service Information

Provider: Baidu
Service Type: API Service
API Endpoint: https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/
Free Type: Permanently Free (some models) + New User Gift
API Compatibility: Supports OpenAI SDK (adjust base_url and api_key)

🎯 Service Overview

Baidu Qianfan LLM Platform provides powerful AI API services, featuring permanently free high-performance models, fully compatible with OpenAI SDK, offering zero-cost AI capabilities for developers.

Core Advantages:

• 🎁 Permanently Free Models - ERNIE-3.5-8K, ERNIE-Speed-8K permanently free
• 💰 Excellent Value - Free models outperform GPT-3.5 Turbo
• 🔄 OpenAI Compatible - Supports OpenAI SDK, one-line migration
• 🇨🇳 Chinese Optimized - Optimized for Chinese, top performance
• 🚀 Fast Domestic Access - Servers in China, quick response
• 🎁 New User Gift - ERNIE-4.0-8K 1M tokens/month (first month)
• 🔒 Compliant for Business - Domestic compliance, safe for commercial use

🚀 Quick Start

Prerequisites

Required:

• ✅ Registered Baidu Cloud account
• ✅ Completed real-name verification
• ✅ Created application and obtained API Key

For detailed steps, see: Baidu Registration Guide

5-Minute Quick Example

Using OpenAI SDK (Recommended)

from openai import OpenAI

# Configure Baidu Qianfan (using OpenAI SDK)
client = OpenAI(
    api_key="YOUR_BAIDU_API_KEY",
    base_url="https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop"
)

# Call permanently free ERNIE-3.5-8K
response = client.chat.completions.create(
    model="ernie-3.5-8k",
    messages=[
        {"role": "system", "content": "You are a helpful assistant"},
        {"role": "user", "content": "Introduce quantum computing"}
    ]
)

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

🤖 Supported Models

Permanently Free Models

New User Gift

Paid Models

⚠️ Important Notes:

• ERNIE-3.5-8K and ERNIE-Speed-8K are permanently free with unlimited tokens
• QPS Limit: Maximum requests per second (can be increased with paid plans)
• New users receive 1M tokens of ERNIE-4.0-8K in first month
• For detailed pricing, see Official Pricing

🔢 Quotas and Limits

Permanently Free Model Limits

New User Gift Limits

Quota Query

• Real-time quota check: Console Quota Page
• Usage statistics: Console → Qianfan Platform → Usage Statistics

💻 Code Examples

1. Basic Conversation (OpenAI SDK)

from openai import OpenAI

client = OpenAI(
    api_key="YOUR_BAIDU_API_KEY",
    base_url="https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop"
)

# Use permanently free ERNIE-3.5
response = client.chat.completions.create(
    model="ernie-3.5-8k",
    messages=[
        {"role": "user", "content": "What is artificial intelligence?"}
    ]
)

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

2. Streaming Output

# Streaming output (for real-time display)
stream = client.chat.completions.create(
    model="ernie-3.5-8k",
    messages=[
        {"role": "user", "content": "Write an article about AI"}
    ],
    stream=True
)

for chunk in stream:
    if chunk.choices[0].delta.content:
        print(chunk.choices[0].delta.content, end="", flush=True)

3. Multi-turn Conversation

messages = [
    {"role": "system", "content": "You are a Python programming assistant"}
]

# First round
messages.append({"role": "user", "content": "How to read CSV files?"})
response = client.chat.completions.create(
    model="ernie-3.5-8k",
    messages=messages
)
messages.append({
    "role": "assistant", 
    "content": response.choices[0].message.content
})

# Second round
messages.append({"role": "user", "content": "How to handle missing values?"})
response = client.chat.completions.create(
    model="ernie-3.5-8k",
    messages=messages
)
print(response.choices[0].message.content)

4. Using Official SDK

# Install official SDK
# pip install qianfan

import qianfan

# Initialize with API Key and Secret Key
chat_comp = qianfan.ChatCompletion(
    api_key="YOUR_API_KEY",
    secret_key="YOUR_SECRET_KEY"
)

# Call ERNIE-3.5-8K
resp = chat_comp.do(
    model="ERNIE-3.5-8K",
    messages=[{
        "role": "user",
        "content": "Hello, please introduce yourself"
    }]
)

print(resp["result"])

5. cURL Example

# Get access_token
curl -X POST \
  'https://aip.baidubce.com/oauth/2.0/token?grant_type=client_credentials&client_id=YOUR_API_KEY&client_secret=YOUR_SECRET_KEY'

# Call API (using obtained access_token)
curl -X POST \
  'https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/ernie-3.5-8k?access_token=YOUR_ACCESS_TOKEN' \
  -H 'Content-Type: application/json' \
  -d '{
    "messages": [
      {
        "role": "user",
        "content": "Hello, please introduce yourself"
      }
    ]
  }'

💡 Best Practices

✅ Recommended Approaches

1. Choose the Right Model

   # General tasks, cost priority → ERNIE-3.5-8K (permanently free)
   model = "ernie-3.5-8k"
   
   # Speed priority → ERNIE-Speed-8K (permanently free)
   model = "ernie-speed-8k"
   
   # Quality priority → ERNIE-4.0-8K (free for new users)
   model = "ernie-4.0-8k"

2. Error Handling and Retry

   import time
   from openai import OpenAI, APIError
   
   def call_with_retry(messages, max_retries=3):
       for i in range(max_retries):
           try:
               return client.chat.completions.create(
                   model="ernie-3.5-8k",
                   messages=messages
               )
           except APIError as e:
               if i < max_retries - 1:
                   wait_time = 2 ** i
                   print(f"API error, waiting {wait_time} seconds...")
                   time.sleep(wait_time)
               else:
                   raise

3. Secure Key Management

   import os
   from dotenv import load_dotenv
   
   load_dotenv()
   api_key = os.getenv('BAIDU_API_KEY')
   secret_key = os.getenv('BAIDU_SECRET_KEY')
   
   client = OpenAI(
       api_key=api_key,
       base_url="https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop"
   )

4. Monitor Usage

   • Regularly visit Console
   • Check remaining quota and usage statistics
   • Set quota alerts

🎯 Optimization Tips

Leverage Permanently Free Models:

• ERNIE-3.5-8K already exceeds GPT-3.5 Turbo
• Suitable for most use cases
• Zero-cost development and operations

Optimize QPS Usage:

• Allocate concurrent requests reasonably
• Excess QPS will queue
• Consider paid upgrade for higher QPS

Result Caching:

import hashlib
import json

cache = {}

def cached_completion(model, messages):
    key = hashlib.md5(
        json.dumps({"model": model, "messages": messages}).encode()
    ).hexdigest()
    
    if key in cache:
        return cache[key]
    
    response = client.chat.completions.create(
        model=model,
        messages=messages
    )
    
    cache[key] = response
    return response

⚠️ Notes

1. Real-name Verification Required: Must complete verification for API use
2. QPS Limits: Free models have QPS of 50, excess will queue
3. Content Moderation: Complies with Chinese laws, sensitive content rejected
4. Token Calculation: Chinese calculated by characters, ~1.5-2 chars = 1 token

🔧 FAQ

1. How to get API Key?

Steps:

1. Login to Baidu Cloud Qianfan Platform
2. Complete real-name verification
3. Create application
4. Get API Key and Secret Key

2. Are permanently free models really unlimited?

Answer: Unlimited tokens! ERNIE-3.5-8K and ERNIE-Speed-8K are permanently free with unlimited token quota, but have QPS (50 requests/second) rate limits. Suitable for small to medium-scale applications.

3. How to use OpenAI SDK?

Answer: Baidu Qianfan supports OpenAI SDK, just modify base_url:

client = OpenAI(
    api_key="YOUR_BAIDU_API_KEY",
    base_url="https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop"
)

4. How does ERNIE-3.5 compare to GPT-3.5?

Answer: According to official data and testing:

• Chinese: ERNIE-3.5 > GPT-3.5 Turbo
• English: ERNIE-3.5 ≈ GPT-3.5 Turbo
• Price: ERNIE-3.5 permanently free, GPT-3.5 paid

5. How to get new user gift?

Answer:

• Register Baidu Cloud account
• Complete real-name verification
• Automatically receive 1M tokens of ERNIE-4.0-8K in first month
• Valid for first month after registration

6. What payment methods are supported?

Answer:

• Alipay
• WeChat Pay
• Bank transfer
• Bank card

📊 Performance Comparison

Comparison with Other APIs

Price Advantage

🌟 Practical Cases

Case 1: AI Customer Service

def ai_customer_service(user_question):
    """AI customer service assistant"""
    response = client.chat.completions.create(
        model="ernie-3.5-8k",  # Use permanently free model
        messages=[
            {
                "role": "system",
                "content": "You are a professional customer service assistant"
            },
            {
                "role": "user",
                "content": user_question
            }
        ]
    )
    return response.choices[0].message.content

# Usage
answer = ai_customer_service("What are the features of your product?")
print(answer)

Case 2: Document Summarization

def summarize_document(document_text):
    """Generate document summary"""
    response = client.chat.completions.create(
        model="ernie-3.5-8k",
        messages=[
            {
                "role": "system",
                "content": "You are a document analysis expert"
            },
            {
                "role": "user",
                "content": f"Please summarize the core content:\n\n{document_text}"
            }
        ]
    )
    return response.choices[0].message.content

Case 3: Content Moderation

def content_moderation(user_content):
    """Content moderation"""
    response = client.chat.completions.create(
        model="ernie-speed-8k",  # Use ultra-fast model
        messages=[
            {
                "role": "system",
                "content": "You are a content moderation expert"
            },
            {
                "role": "user",
                "content": f"Judge if this content is compliant:\n{user_content}"
            }
        ]
    )
    return response.choices[0].message.content

📚 Related Resources

Official Documentation

• Complete Baidu Introduction
• Chatbot Usage Guide
• API Official Documentation

Tools and Resources

• Qianfan Console
• Quota Query
• Pricing
• Real-name Verification
• Official SDK (Python)

📝 Update Log

• 2026-01-28: Document verification and optimization, clarified quota limits
• 2025: ERNIE-3.5-8K and ERNIE-Speed-8K permanently free
• March 2025: Released ERNIE 4.5 and ERNIE X1 new models
• Continuous Updates: Ongoing optimization of API performance and stability

Service Provider: Baidu