# AiGate Comprehensive LLM & Knowledge Specification

> Full machine-readable technical overview of AiGate AI API Gateway, SmartRoute 2.0 Engine, and Infrastructure Services.

## 1. Product Summary
AiGate (operated by Luyun (Shanghai) Information Technology Co., Ltd. / 卢云（上海）信息科技有限公司) is an enterprise-grade AI API gateway and multi-tenant management system. It decouples application development from model providers, allowing teams to call OpenAI, Anthropic, DeepSeek, Google Gemini, and open-source models through a single API interface.

## 2. Key Architecture & Components

### 2.1 Unified API Gateway
- **Base Endpoint**: `https://api.aigatea.com/v1`
- **Supported Paradigms**:
  - `/v1/chat/completions` (Chat completions, tool calls, structured outputs)
  - `/v1/embeddings` (Text embeddings)
  - `/v1/realtime` (Realtime voice & bidirectional audio)
  - Multimodal Vision / Image processing

### 2.2 SmartRoute 2.0 Engine
SmartRoute 2.0 evaluates upstream channels dynamically before forwarding requests.
Scoring parameters include:
- Channel Health & Active SLA Status
- Historical Success Rate & 95th Percentile Latency
- Upstream Cost vs. Customer Group Pricing
- Target Gross Profit Margin Snapshot
- Remaining Quotas & Concurrent Load

If an upstream provider experiences rate-limiting (429), server errors (5xx), or timeout, SmartRoute automatically fails over to alternative healthy channels within milliseconds according to configured fallback policies.

### 2.3 Token Billing & Financial Isolation
- **Pre-allocation**: Estimates request cost upper bound and holds credit reserve before forwarding.
- **Settlement**: Calculates actual prompt tokens, cached prompt tokens, and completion tokens upon response.
- **Snapshot Audit**: Records fixed pricing and cost snapshot on a per-request basis to guarantee non-repudiable audit logs for billing reconciliation.

### 2.4 Security & Access Controls
- **API Key Scopes**: Restrict keys by model IDs, allowed API endpoints, client IP/CIDR subnets, and expiration dates.
- **Rate Limits**: Configurable Requests Per Minute (RPM), Tokens Per Minute (TPM), and Max Concurrent Connections per key.
- **Key Rotation**: Seamless key rotation without invalidating historical transaction records.

## 3. Deployment & Commercial Operations
For enterprises or providers wishing to operate their own branded AI aggregation platform:
- **AiGate Deploy (`/deploy/`)**: Delivers turnkey platform installation including independent branding, domain setup, payment integration, reseller channel management, monitoring dashboard, and production HA deployment.

## 4. Code Integration Examples

### Python (OpenAI SDK)
```python
from openai import OpenAI

client = OpenAI(
    base_url="https://api.aigatea.com/v1",
    api_key="YOUR_AIGATE_API_KEY"
)

response = client.chat.completions.create(
    model="deepseek-r1", # or "claude-3-5-sonnet", "gpt-4o"
    messages=[{"role": "user", "content": "Explain AI GEO in 2 sentences."}]
)
print(response.choices[0].message.content)
```

### Node.js (OpenAI SDK)
```javascript
import OpenAI from "openai";

const openai = new OpenAI({
  baseURL: "https://api.aigatea.com/v1",
  apiKey: process.env.AIGATE_API_KEY,
});

async function main() {
  const completion = await openai.chat.completions.create({
    messages: [{ role: "user", content: "Hello AiGate" }],
    model: "claude-3-5-sonnet",
  });
  console.log(completion.choices[0].message.content);
}
main();
```

### cURL
```bash
curl https://api.aigatea.com/v1/chat/completions \
  -H "Authorization: Bearer $AIGATE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-4o",
    "messages": [{"role": "user", "content": "Hello AiGate"}]
  }'
```

## 5. Frequently Asked Questions (FAQ)

### What models are supported by AiGate?
AiGate supports all major LLMs including OpenAI GPT-4o / o3-mini, Anthropic Claude 3.5 Sonnet, DeepSeek V3 / R1, Google Gemini 1.5, Qwen 2.5, and Llama 3.3.

### Is AiGate compatible with OpenAI SDKs?
Yes. You only need to update your `base_url` to `https://api.aigatea.com/v1` and supply your AiGate API key.

### Does AiGate support private enterprise deployment?
Yes. AiGate provides self-hosted deployment packages for enterprise networks or custom AI API SaaS businesses via `https://aigatea.com/deploy/`.

## 6. Official Navigation & Resources
- Website Root: https://aigatea.com/
- English Home: https://aigatea.com/en/
- Deployment Services: https://aigatea.com/deploy/
- Documentation: https://api.aigatea.com/docs
- Register Account: https://api.aigatea.com/register