Why I built luoluoAI
I started building luoluoAI after my own AI development workflow became increasingly fragmented.
I use AI APIs for coding tools, automation, testing, and small applications. Using official providers directly worked well, but maintaining several integrations meant managing separate API keys, balances, dashboards, request formats, and usage limits.
The subscriptions did not completely solve this problem either. Chat subscriptions are useful for interactive work, but scripts, agents, and applications still require separately billed API access.
I also learned that “OpenAI-compatible” does not always mean behavior-compatible. Basic chat requests are usually straightforward, but differences become visible with streaming responses, fragmented tool calls, content blocks, usage metadata, errors, and retries.
I built luoluoAI as an independent API gateway to simplify this workflow. It currently provides:
- OpenAI-compatible API access
- Claude Messages-compatible requests
- streaming responses
- model configuration
- usage and balance tracking
- Python, Node.js, and cURL examples
Keeping the service affordable is important to me. Pricing is published by model so developers can compare costs before integrating, rather than discovering the effective price after generating significant traffic.
This is still an evolving independent project. My current priorities are improving compatibility, making failures easier to diagnose, and keeping pricing and usage records transparent.
Documentation:
https://api.luoluocoder.com
Model pricing:
https://api.luoluocoder.com/model-plaza
Example integrations:
https://github.com/siofeng/openai-compatible-api-examples
I maintain luoluoAI and would particularly value feedback about streaming behavior, tool calls, documentation, model coverage, and the onboarding experience.
