7.5k stars! This open source API gateway completely liberates developers: one-click aggregation of GPT-4, Suno, Midjourney, and online recharge!
An open source API gateway that makes AI model calls and payments easier than ever before.
Core content:
1. Multi-model unified gateway, supporting AI capabilities such as GPT-4 and Suno music generation
2. Enterprise-level permission management and original reasoning strength classification, balancing costs and effects
3. Magic-free access to global AI services, supporting Docker one-click deployment and flexible billing system
New API is an AI model interface management and distribution system based on One API secondary development . It supports the unified packaging of multiple large models (such as GPT-4, Suno, Midjourney, etc.) into OpenAI format interface calls, and is compatible with the Easy Payment Protocol. Whether it is an individual developer or an enterprise, it can use it to efficiently manage AI assets and reduce development costs.
Core Features
1. Multi-model unified gateway, say goodbye to interface fragmentation
It supports multiple AI capabilities such as GPT-4, Suno music generation, Midjourney drawing agent, Rerank sorting model, etc., which can be called uniformly through standardized interfaces. Developers do not need to connect to each model separately, and can switch freely with just one key.
2. Enterprise-level permission management
Model access control : assign callable model permissions by user/role Intelligent routing strategy : supports weighted random allocation of channels to avoid overloading of a single interface Real-time monitoring dashboard : data visualization displays core indicators such as interface call volume and success rate
3. Original “Degree of Reasoning” grading
Add a suffix to the model name (such as-high/-medium/-low), you can dynamically adjust the computing power consumption. Users can balance cost and effect according to scenario requirements, for example:
# High computing power mode (accurate but time-consuming)
model = "gpt-4-high"
# Low computing power mode (fast response)
model = "gpt-4-low"
4. No-magic access to global AI services
By deploying an overseas server proxy, you can call OpenAI and other restricted interfaces without complex network configuration. The actual deployment only requires 3 steps:
# 1. Docker one-click deployment
docker run --name new-api -d -p 3000:3000 calciumion/new-api:latest
# 2. Nginx reverse proxy (sample configuration)
server {
listen 80;
server_name api.yourdomain.com;
location / {
proxy_pass http://localhost:3000;
}
}
# 3. HTTPS certificate automation (Let's Encrypt)
sudo certbot --nginx
5. Flexible billing system
Pay-per-view : Set independent unit prices for different models Online recharge : integrated Alipay/WeChat payment interface Credit limit warning : set usage thresholds for automatic reminders
6. Developer-friendly design
Telegram login : password-free, fast access Multi-language interface : free switching between Chinese and English Docker multi-machine deployment : support Redis cluster cache and session synchronization
Technical Architecture
| Interface Gateway | ||
| Session Management | ||
| Deployment and Operation | ||
| Safety protection | ||
| Data storage |
Four typical application scenarios
Scenario 1: Enterprise internal AI platform
An e-commerce company uses the New API to centrally manage three major AI services: customer service robots, product copy generation, and image review, increasing development efficiency by 70%.
Scenario 2: Multi-model SaaS platform
Developer @TechBot uses weighted random routing to intelligently distribute user requests to GPT-4 and Claude-3, reducing costs by 40% while maintaining 99% SLA.
Scenario 3: Academic Research Collaboration
Through decentralized management, university laboratories allow graduate students to call Stable Diffusion and LLAMA2 on demand, doubling the efficiency of paper production.
Scenario 4: Personal Developer Tools
Independent developer "AI Xiaojiang" combined the online recharge function and built a paid question-and-answer robot in 3 days, with monthly turnover exceeding 50,000 yuan.
Interface effects at a glance
Comparison of similar projects
| Multiple model support | |||
| Online Payment | |||
| Reasoning Strength Grading | |||
| Deployment complexity | |||
| Enterprise-grade features |
Summarize
New API is becoming the new favorite of AI developers with its features of "out-of-the-box, comprehensive functions and controllable costs" . Whether it is an individual developer who wants to quickly verify the idea or a technical team that needs to build an enterprise-level AI middle platform, they can get the ultimate experience from it.
