Lurus API — Privately Deployed Multi-Tenant LLM Gateway
A multi-tenant LLM gateway deployed in the customer's own environment · Based on the open-source New API, AGPLv3
One API KeyAPI KeyA secret credential for accessing the API, much like a password. Each Key can have its own quota and permissions.Learn more → to access 50+ leading AI models. Fully compatible with the OpenAI SDKSDKA software development kit providing ready-made functions and classes to simplify API calls. The Lurus gateway uses the common Chat Completions interface format. — your existing code needs only a two-line change, no rewrite required.
Which kind of user are you?
Three entry points — pick the one that matches your background and start right away.
Set up an AI client first (Cherry Studio / Lobe Chat), enter your API Key, and start chatting — no code required at any point.
Make your first API call in 5 minutes — supports Python / Node.js / Go / cURL.
Migrate by replacing two lines of code — all OpenAI SDK features are fully compatible.
Already using OpenAI, want to switch / cut costs
Migrate by replacing two lines of code — all OpenAI SDK features are fully compatible:
# 改这两行，其余代码不动
base_url="https://hub.lurus.cn/v1"
api_key="sk-your-lurus-key"One gateway, four jobs
Unified access, intelligent routing, cost control, and enterprise-grade access management.
Unified API
One interface covers every model — just change the model name, no need to write an adapter for each provider.
Intelligent Routing & Automatic Failover
Multi-channel backup (automatically switches when the primary channel fails), weighted load balancing (proportional traffic split to balance cost and speed), and priority policies (low-cost channels first, switching to higher-cost backups when limits are exceeded).
Fine-grained Cost Control
Set a Token quota per API Key with over-limit blocking; review call counts, Tokens, and cost breakdowns by day/month; get alerts when remaining quota drops below 20%.
Enterprise-grade Access Management
Allocate multiple Keys per project, model allowlists, IP allowlists, and complete audit logs — every request records the model, Tokens, and latency.
Unified API example — switch providers by changing the model name:
from openai import OpenAI
client = OpenAI(base_url="https://hub.lurus.cn/v1", api_key="sk-your-api-key")
# model 可填 deepseek-chat / gpt-4o / claude-3-5-sonnet / gemini-3-pro-preview
response = client.chat.completions.create(model="deepseek-chat", messages=[{"role": "user", "content": "你好"}])Enterprise-grade Access Management
| Feature | Description |
|---|---|
| Multi-Key management | Assign independent Keys to different projects/teams |
| Model allowlist | Restrict a Key to specified models only |
| IP allowlist | Allow calls only from specified IP ranges |
| Complete audit logs | Every request records the model, Tokens, and latency |
Who uses Lurus API
| Scenario | What you can do |
|---|---|
| AI application development | Use the same code to access all providers, and quickly A/B test different models |
| Cost optimization | Route everyday tasks to DeepSeek (low cost) and complex tasks to GPT-4o (high quality) |
| Service reliability | Multi-channel redundancy — a single provider outage won’t affect your service |
| Team management | Allocate Keys + quotas, and review everyone’s AI usage and costs in one place |
| AI clients | Provide a unified backend for tools like Cherry Studio, Lobe Chat, and OpenCat |
How a request flows
graph LR; A[Your App / AI Client] --> B[Lurus API Gateway]; B --> C[Auth]; C --> D[Routing]; D --> E[Rate Limiting]; E --> F[Billing]; F --> G[Logging]; D --> H[OpenAI]; D --> I[Claude]; D --> J[Gemini]; D --> K[DeepSeek]The gateway routes by the configured channel priority; when a provider returns an error it automatically retries the next one, and your code never sees the switch.
Run through the whole flow in 20 minutes
First time here? Follow the order.
Get an API Key — sign up and create your first Key (3 minutes)
Quickstart — make your first API request (5 minutes)
Supported Models — learn which models are available and how to choose
Chat Completions API — master the most commonly used endpoint
Advanced users, jump straight to…
- Function Calling — let the AI call your functions
- Streaming Responses — token-by-token output for a better experience
- API Reference Overview — the complete endpoint list