DragonAPI — Access China's Best AI Models

One API key. OpenAI-compatible. No Chinese phone number needed.

DragonAPI is an API gateway that gives global developers instant access to China's top AI models — DeepSeek, Qwen, GLM, and more — through a standard OpenAI-compatible interface. Just change your base_url and you're up and running.

Why? Models like DeepSeek V3 and Qwen 3 rival GPT-4o at a fraction of the cost, but accessing them from outside China requires a Chinese phone number, WeChat Pay, and navigating Chinese-only interfaces. DragonAPI removes all those barriers.

---

Why Developers Choose DragonAPI

	DragonAPI	Direct China API	OpenAI
Chinese phone number required	No	Yes	No
WeChat/Alipay required	No	Yes	No
OpenAI SDK compatible	Yes	No	Yes
DeepSeek V3 pricing (per 1M input tokens)	$0.13	$0.27	N/A
DeepSeek R1 pricing (per 1M input tokens)	$0.28	$0.55	N/A
GPT-4o equivalent pricing	~90% cheaper	N/A	$5.00
Credit card payment	Yes	No	Yes
Latency (US endpoint)	~200ms TTFT	High (from outside China)	~150ms

---

Quick Start

1. Get your API key

Sign up at the DragonAPI Dashboard and create an API key.

2. Use with any OpenAI SDK

Python

from openai import OpenAI

client = OpenAI(
    api_key="sk-your-dragonapi-key",
    base_url="http://154.21.86.24:3000/v1"
)

response = client.chat.completions.create(
    model="deepseek-chat",  # DeepSeek V3
    messages=[{"role": "user", "content": "Explain quantum computing in simple terms"}]
)
print(response.choices[0].message.content)

JavaScript / TypeScript

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: "sk-your-dragonapi-key",
  baseURL: "http://154.21.86.24:3000/v1",
});

const response = await client.chat.completions.create({
  model: "deepseek-chat",
  messages: [{ role: "user", content: "Write a Python function to sort a list" }],
});
console.log(response.choices[0].message.content);

cURL

curl http://154.21.86.24:3000/v1/chat/completions \
  -H "Authorization: Bearer sk-your-dragonapi-key" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek-chat",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

---

Supported Models

DeepSeek

Model	Best For	Input (per 1M tokens)	Output (per 1M tokens)
deepseek-chat	General chat, coding, writing	$0.13	$0.28
deepseek-coder	Code generation & completion	$0.13	$0.28
deepseek-reasoner	Complex reasoning (R1)	$0.28	$0.55

Qwen (Alibaba)

Model	Best For	Input (per 1M tokens)	Output (per 1M tokens)
qwen-turbo	Fast, cheap inference	$0.05	$0.20
qwen-plus	Balanced quality & cost	$0.13	$0.78

Route Aliases (Smart Routing)

Use simple aliases and we'll route to the best model for the job:

Route	Maps To	Best For	Input	Output
fast	Qwen Turbo	Simple chat, classification	$0.05	$0.20
balanced	DeepSeek Chat	SaaS chat, summarization	$0.13	$0.78
coding	Qwen Coder Plus	Code agents, generation	$0.05	$0.20
reasoning	DeepSeek R1	Analysis, reasoning	$0.30	$1.04

Additional Models

GLM-4, Yi, SparkDesk, Hunyuan, and more — see the full list on the pricing page.

---

Supported API Formats

DragonAPI isn't limited to OpenAI format. We support multiple API formats through a single endpoint:

Format	Endpoint	Use Case
OpenAI Chat	POST /v1/chat/completions	Default, most SDKs
OpenAI Responses	POST /v1/responses	New OpenAI responses API
Anthropic / Claude	POST /v1/messages	Claude SDK compatibility
Google Gemini	POST /v1beta/models/*	Gemini SDK compatibility
Embeddings	POST /v1/embeddings	Text embeddings
Image Generation	POST /v1/images/generations	DALL-E compatible
Audio	POST /v1/audio/speech	TTS & transcription
Rerank	POST /v1/rerank	Search reranking

---

Framework Integration

DragonAPI works out of the box with popular AI frameworks:

LangChain

from langchain_openai import ChatOpenAI

llm = ChatOpenAI(
    model="deepseek-chat",
    openai_api_key="sk-your-dragonapi-key",
    openai_api_base="http://154.21.86.24:3000/v1"
)

response = llm.invoke("Summarize the key points of transformer architecture")
print(response.content)

LlamaIndex

from llama_index.llms.openai import OpenAI

llm = OpenAI(
    model="deepseek-chat",
    api_key="sk-your-dragonapi-key",
    api_base="http://154.21.86.24:3000/v1"
)

response = llm.complete("Explain RAG in 3 sentences")
print(response)

Vercel AI SDK

import { createOpenAI } from "@ai-sdk/openai";
import { generateText } from "ai";

const dragonapi = createOpenAI({
  apiKey: "sk-your-dragonapi-key",
  baseURL: "http://154.21.86.24:3000/v1",
});

const { text } = await generateText({
  model: dragonapi("deepseek-chat"),
  prompt: "Write a haiku about programming",
});

AutoGen / CrewAI

# Works with any framework that accepts OpenAI-compatible config
config = {
    "model": "deepseek-chat",
    "api_key": "sk-your-dragonapi-key",
    "base_url": "http://154.21.86.24:3000/v1"
}

---

Cost Comparison

How much can you save by switching from OpenAI?

Use Case	OpenAI (GPT-4o)	DragonAPI (DeepSeek V3)	Savings
1M input tokens	$5.00	$0.13	97%
1M output tokens	$15.00	$0.28	98%
10k API calls/day (avg 500 tokens each)	~$75/day	~$2/day	$73/day
Monthly (moderate SaaS)	~$2,250	~$60	$2,190/mo

---

Features

• OpenAI SDK drop-in replacement — Change one line of code (base_url) and everything works
• Pay-as-you-go — No monthly fees, no commitments, no minimum spend
• Credit card payments — No WeChat Pay or Alipay required
• US & CN endpoints — Low latency wherever your servers are
• Streaming support — Full SSE streaming for real-time responses
• Multi-format — OpenAI, Claude, Gemini formats all supported
• Smart routing — Use aliases like fast, balanced, coding, reasoning
• Image & video generation — DALL-E compatible image gen, plus Sora/Kling video gen
• Audio — TTS and speech-to-text support

---

API Reference

Full OpenAPI documentation is available at the API Docs.

Authentication

All requests require a Bearer token:

Authorization: Bearer sk-your-dragonapi-key

Base URLs

Endpoint	URL	Best For
US (recommended)	http://154.21.86.24:3000/v1	Users in Americas/Europe
CN	http://8.130.88.75:3000/v1	Users in Asia-Pacific

---

FAQ

Is this legal?

Yes. We relay API requests to Chinese AI providers. The models themselves are publicly available commercial APIs. We simply handle the payment and access infrastructure so you don't need Chinese credentials.

What about data privacy?

Your API requests are relayed to the upstream model providers (DeepSeek, Alibaba Cloud, etc.). We do not store your prompts or responses. The same data handling policies of the upstream providers apply.

How is latency?

The US endpoint is hosted in the United States. Typical TTFT (time to first token) is ~200ms for DeepSeek models. Streaming is fully supported.

Can I use this in production?

Yes. Many developers use DragonAPI for production SaaS applications. We recommend using the route aliases (fast, balanced, coding, reasoning) for automatic fallback.

What if a model goes down?

Smart routing aliases automatically fail over to the best available model in that category.

---

Get Started

1. Sign up at DragonAPI Dashboard
2. Create an API key in the dashboard
3. Replace your base_url — that's it, you're done

# Before (OpenAI — $5/1M input tokens)
client = OpenAI(api_key="sk-openai-key")

# After (DragonAPI — $0.13/1M input tokens)
client = OpenAI(
    api_key="sk-dragonapi-key",
    base_url="http://154.21.86.24:3000/v1"
)

---

License

MIT

---

Built for developers who want the best AI models at the best prices.
Dashboard · Pricing · API Docs