Fully OpenAI Compatible
Just change base_url and api_key to seamlessly connect any OpenAI SDK or HTTP client without modifying a single line of business code.
Fully OpenAI Compatible · Multi-Model Routing
One Line of CodeAccess Domestic LLM APIs
Access DeepSeek, Zhipu, Qwen, Doubao and more in one place. High-availability relay, real-time billing, developer-friendly.
quickstart.py
from openai import OpenAI
client = OpenAI(
api_key="YOUR_API_KEY",
base_url="https://www.ciyuano.com/v1"
)
response = client.chat.completions.create(
model="deepseek-v4-flash",
messages=[{"role": "user", "content": "Hello!"}]
)
print(response.choices[0].message.content)Compatible with
and more
and moreWe Are Growing Fast
12+
Available Models
4
Upstream Providers
99.9%
Service Uptime
<200ms
Avg Latency
Supported Models
Access All Major LLMs in One Place
Covering DeepSeek, Zhipu, Qwen, Doubao and more, constantly expanding
Xiaomi MiMo
2 Models
Agnes
6 ModelsAgnes 2.0 Flash
agnes-2.0-flashInput ¥0/1M·Output ¥0/1M
Context 125K tokens
Agnes Image 2.0 Flash
agnes-image-2.0-flashInput ¥0/1M·Output ¥0/1M
Context 125K tokens
Agnes Image 2.5 Flash
agnes-image-2.5-flashInput ¥2/1M·Output ¥8/1M
Context 125K tokens
Agnes-Video-V2.0
agnes-video-v2.0Input ¥2/1M·Output ¥8/1M
Context 125K tokens
agnes-2.5-flash
agnes-2.5-flashInput ¥2/1M·Output ¥8/1M
Context 125K tokens
agnes-2.5-pro
agnes-2.5-proInput ¥2/1M·Output ¥8/1M
Context 125K tokens
Core Advantages
Built for Developers
One-stop AI API solution, integrating LLMs is as easy as breathing
Multi-Model Smart Routing
Support DeepSeek, GLM, Qwen and more with intelligent load balancing and automatic failover.
High Availability
Channel health checks, key pool rotation, automatic retries ensuring 99.9% API uptime.
Real-time Billing
Precise per-request token billing with visual dashboard for real-time usage and cost monitoring.
Streaming Response
Full SSE streaming support with real-time reasoning model thinking process, native-like experience.
Secure & Controllable
Independent API Key system, balance control, model whitelist, request log audit.
Quick Start
Get Started in 3 Steps
No code changes needed, migrate instantly
1
Register
Free registration with instant trial credits
2
Create API Key
Generate keys with one click in the dashboard
3
Integrate
Change base_url and start calling
Python
from openai import OpenAI
client = OpenAI(
api_key="YOUR_API_KEY",
base_url="https://www.ciyuano.com/v1"
)
response = client.chat.completions.create(
model="deepseek-v4-flash",
messages=[{"role": "user",
"content": "Hello!"}]
)Transparent Pricing
Simple, Pay-as-you-go Pricing
No monthly fees. Pay only for what you use. Free credits on registration.
| Provider | Model | Input | Output | Context |
|---|---|---|---|---|
| agnes | agnes-2.0-flash | ¥0 | ¥0 | 125K |
| agnes | agnes-image-2.0-flash | ¥0 | ¥0 | 125K |
| xiaomi | mimo-v2.5 | ¥1 | ¥2 | 1024K |
| agnes | agnes-image-2.5-flash | ¥2 | ¥8 | 125K |
| agnes | agnes-video-v2.0 | ¥2 | ¥8 | 125K |
 sensenova | deepseek-v4-flash | ¥2 | ¥8 | 125K |
| sensenova | glm-5.2 | ¥2 | ¥8 | 125K |
| sensenova | sensenova-6.8-flash-lite | ¥2 | ¥8 | 125K |
Comparison
Why Not Go Direct?
One API key, all models. No need to manage multiple vendor accounts.
| Feature | Ciyuano | Direct API | Other Relay |
|---|---|---|---|
| OpenAI Compatible | |||
| Multi-model Support | — | ||
| Auto Failover | — | — | |
| Key Pool Rotation | ✗ | — | |
| Smart Routing | ✗ | ✗ | |
| Usage Dashboard | ✗ | ✗ | |
| Free Trial Credits | ✗ | ✗ |
Blog
Latest Articles
Tech sharing and product updates
Productivity Tools
·2 min read·354 views
AI Resume Writing: From Zero to Professional in 5 Steps
Learn how to use AI to write a professional resume. Five simple steps from basic information to polished final draft. Perfect for job seekers who want to stand out.
Read More
Tutorials
·2 min read·416 views
AI Speech to Text: Beginner's Guide to Convert Voice to Text in 3 Steps
Learn how to convert speech to text using AI tools. Perfect for meeting notes, content creation, and video subtitles. Three practical scenarios with step-by-step instructions.
Read More
Tech Frontier
·2 min read·395 views
AI Prompting Tips: 7 Practical Methods for More Accurate Answers
Learn 7 practical prompting techniques to get more accurate and useful AI responses. From role assignment to step-by-step instructions, master the art of asking AI the right way.
Read More
Ready to Get Started?
Register for free trial credits, complete integration in 5 minutes. Supports DeepSeek, GLM, Qwen and all major models.