📅 July 24, 2026 · 📖 10 min read · Last updated: July 2026

OpenAI Compatible API:
How to Switch in 5 Minutes

You've been using OpenAI's API. The pricing is climbing. You've heard about cheaper alternatives. But rewriting your entire codebase? That's a non-starter.

Good news: you don't need to rewrite anything. Dozens of world-class LLM providers now offer OpenAI-compatible APIs. You change two lines of code — the base URL and API key — and you're running on models that are 80–97% cheaper.

1. Why You Need an OpenAI Compatible API

OpenAI's API has become the de facto standard for LLM interactions. But relying on a single provider creates risks:

• Cost escalation: GPT-4o at $2.50/$10.00 per million tokens is expensive at scale
• Vendor lock-in: All your logic, prompts, and infrastructure tied to one provider
• Rate limits: OpenAI throttles during peak demand — your app goes down
• Model availability: When GPT-4o has issues, you have no fallback

An OpenAI compatible API solves all of these. You keep the same SDK, same code patterns, same mental model — but gain access to 38+ models from multiple providers, at dramatically lower costs.

2. What "OpenAI Compatible" Actually Means

An API is "OpenAI compatible" when it implements the same HTTP endpoints, request format, and response format as OpenAI's API. Specifically:

In practice, this means: if your code works with OpenAI, it works with any compatible provider. Period.

3. The Migration: Step by Step

What changes in your code:

4. Complete Code Examples

Python — Full Migration Example

from openai import OpenAI

# ============================================
# BEFORE: OpenAI
# ============================================
# client = OpenAI(api_key="sk-your-openai-key")
# response = client.chat.completions.create(
#     model="gpt-4o",
#     messages=[{"role": "user", "content": "Hello!"}]
# )

# ============================================
# AFTER: AI Token (2 lines changed!)
# ============================================
client = OpenAI(
    base_url="http://47.237.87.92:3001/v1",
    api_key="your-ai-token-key"
)

# DeepSeek V4 Flash — 97% cheaper than GPT-4o
response = client.chat.completions.create(
    model="deepseek-ai/DeepSeek-V4-Flash",
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "Hello!"}
    ],
    temperature=0.7,
    max_tokens=1024
)
print(response.choices[0].message.content)


# ============================================
# Streaming works too — no changes needed!
# ============================================
stream = client.chat.completions.create(
    model="qwen/Qwen2.5-72B-Instruct",
    messages=[{"role": "user", "content": "Write a poem about code"}],
    stream=True
)
for chunk in stream:
    if chunk.choices[0].delta.content:
        print(chunk.choices[0].delta.content, end="")


# ============================================
# Function calling works too!
# ============================================
import json

tools = [
    {
        "type": "function",
        "function": {
            "name": "get_weather",
            "description": "Get weather for a location",
            "parameters": {
                "type": "object",
                "properties": {
                    "location": {"type": "string"},
                    "unit": {"type": "string", "enum": ["celsius", "fahrenheit"]}
                },
                "required": ["location"]
            }
        }
    }
]

response = client.chat.completions.create(
    model="deepseek-ai/DeepSeek-V4-Pro",
    messages=[{"role": "user", "content": "What's the weather in Tokyo?"}],
    tools=tools
)
print(response.choices[0].message.tool_calls)

Node.js — Full Migration Example

import OpenAI from 'openai';

// BEFORE: OpenAI
// const client = new OpenAI({ apiKey: 'sk-your-openai-key' });

// AFTER: AI Token — just change baseURL and apiKey!
const client = new OpenAI({
  baseURL: 'http://47.237.87.92:3001/v1',
  apiKey: 'your-ai-token-key'
});

// Basic chat completion
const response = await client.chat.completions.create({
  model: 'deepseek-ai/DeepSeek-V4-Flash',
  messages: [
    { role: 'system', content: 'You are a helpful assistant.' },
    { role: 'user', content: 'Explain recursion in one paragraph.' }
  ]
});
console.log(response.choices[0].message.content);

// Streaming
const stream = await client.chat.completions.create({
  model: 'qwen/Qwen2.5-72B-Instruct',
  messages: [{ role: 'user', content: 'Write a haiku about APIs' }],
  stream: true
});
for await (const chunk of stream) {
  process.stdout.write(chunk.choices[0]?.delta?.content || '');
}

cURL

# Same format as OpenAI — just different base URL and model
curl http://47.237.87.92:3001/v1/chat/completions \
  -H "Authorization: Bearer your-ai-token-key" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek-ai/DeepSeek-V4-Flash",
    "messages": [
      {"role": "system", "content": "You are a concise assistant."},
      {"role": "user", "content": "What is 2+2?"}
    ],
    "temperature": 0.7,
    "max_tokens": 100,
    "stream": false
  }'

5. Available Models Through AI Token

Once connected, you can use any of these models by changing the model parameter:

6. Advanced: Smart Routing

With AI Token, you can implement intelligent model routing — using the cheapest model that's good enough for each task:

from openai import OpenAI

client = OpenAI(
    base_url="http://47.237.87.92:3001/v1",
    api_key="your-ai-token-key"
)

def smart_complete(messages, task_type="general"):
    """Route to the best model based on task type."""
    model_map = {
        "general":    "deepseek-ai/DeepSeek-V4-Flash",   # $0.14/M — fast & cheap
        "chinese":    "qwen/Qwen2.5-72B-Instruct",       # $0.15/M — best Chinese
        "reasoning":  "deepseek-ai/DeepSeek-R1",         # $0.55/M — deep thinking
        "long_doc":   "moonshot/kimi-128k",              # $0.20/M — 128K context
        "free":       "zhipu/glm-4-flash",               # $0.00 — completely free
        "premium":    "openai/gpt-4o",                   # $2.50/M — frontier quality
    }
    
    model = model_map.get(task_type, model_map["general"])
    
    return client.chat.completions.create(
        model=model,
        messages=messages,
        temperature=0.7
    )

# Usage — just specify the task type!
result = smart_complete(
    messages=[{"role": "user", "content": "Summarize this meeting transcript..."}],
    task_type="general"
)

# Chinese content — auto-routed to Qwen
result = smart_complete(
    messages=[{"role": "user", "content": "帮我翻译这段话成英文..."}],
    task_type="chinese"
)

7. Common Migration Concerns

8. Migration Checklist

1. ☐ Get your AI Token API key from /pay/starter
2. ☐ Update base_url in your OpenAI client initialization
3. ☐ Update api_key to your new key
4. ☐ Change model names to your preferred alternatives
5. ☐ Test with the free playground first
6. ☐ Run your test suite to verify responses
7. ☐ Deploy to production 🚀

FAQ

Related Resources

• LLM Playground — Test all models before switching
• DeepSeek vs OpenAI: Complete Comparison
• Complete Guide to Chinese LLM APIs
• API Documentation
• Model Comparison Table