# NextRouter AI Token (新词元 AI Token)

> AI Model Token Gateway Platform — unified OpenAI-compatible API for all major large language models.

NextRouter AI Token is an enterprise-grade AI model token routing and management platform. It aggregates multiple upstream AI providers (OpenAI, Anthropic, DeepSeek, Google Gemini, Qwen, Zhipu AI, MiniMax) behind a single OpenAI-compatible API endpoint, enabling developers and businesses to access all major LLMs with one API key.

## Core Capabilities

- **OpenAI-Compatible API**: Drop-in replacement for OpenAI SDK. Supports `/v1/chat/completions`, `/v1/embeddings`, `/v1/models`, `/v1/rerank`, `/v1/images/generations`, `/v1/audio/speech`.
- **Multi-Provider Aggregation**: Unified access to 7+ providers and 30+ models including GPT-4o, Claude, DeepSeek-V3, Qwen-Max, GLM-4, and more.
- **Multi-Channel Load Balancing**: Weighted routing across multiple upstream channels per model, with priority-based fallback.
- **Circuit Breaker & Auto-Recovery**: Automatic failure detection on upstream channels with 30-second circuit breaking and intelligent failover.
- **Token-Level Billing**: Precise per-token metering with prepaid wallet system, redeem codes, and online payment (Alipay).
- **Multi-Tenant Architecture**: Platform → Agent → Company → Employee four-tier hierarchy with RBAC data isolation.
- **API Key Management**: Multiple keys per account with token group multipliers, daily limits, rate limiting, and IP allowlists.
- **Real-Time Analytics**: Usage monitoring, trend analysis, multi-dimensional rankings powered by ClickHouse.
- **Security**: IP blacklist, TOTP 2FA, login lockout, request rate limiting.
- **Agent/Reseller System**: Agents can set custom pricing multipliers, manage downstream companies, and issue redeem codes.

## Supported Providers & Models

| Provider | Example Models |
|----------|---------------|
| OpenAI | GPT-4o, GPT-4o-mini, o1, o3-mini, text-embedding-3 |
| Anthropic | Claude 3.5 Sonnet, Claude 3.5 Haiku |
| DeepSeek | DeepSeek-V3, DeepSeek-R1 |
| Google | Gemini 2.0 Flash, Gemini 2.0 Flash-Lite |
| Qwen (Alibaba) | Qwen-Max, Qwen-Plus, Qwen-Turbo |
| Zhipu AI | GLM-4-Plus, GLM-4-Flash |
| MiniMax | MiniMax-Text-01 |

## Quick Start

```bash
# Use with any OpenAI-compatible SDK
curl https://token.bitbeam.cn/v1/chat/completions \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "gpt-4o", "messages": [{"role": "user", "content": "Hello"}]}'
```

## Pricing

Token-based billing (per million tokens) with transparent pricing displayed in the Model Market. Agents receive wholesale rates with configurable markup multipliers.

## Links

- Homepage: https://token.bitbeam.cn
- Model Market: https://token.bitbeam.cn/model-market
- API Documentation: https://token.bitbeam.cn/docs
- Become an Agent: https://token.bitbeam.cn/apply-agent
- Blog: https://token.bitbeam.cn/blog
- Register: https://token.bitbeam.cn/register

## Frequently Asked Questions

**Q: What is NextRouter AI Token?**
A: NextRouter AI Token is an AI model token gateway platform that provides a unified OpenAI-compatible API. Access 30+ models from 7+ providers including OpenAI, Anthropic, DeepSeek, Google Gemini, Qwen, Zhipu AI, and MiniMax through a single endpoint.

**Q: How do I get started? Do I need to change my code?**
A: Fully compatible with OpenAI SDK (Python / Node.js / Go / Java). Just replace the base_url with https://token.bitbeam.cn/v1 and api_key with your key. No other code changes needed.

**Q: Which AI models are supported?**
A: Supports GPT-4o, GPT-4o-mini, o1, o3-mini, Claude 3.5 Sonnet/Haiku, DeepSeek-V3, DeepSeek-R1, Gemini 2.0 Flash, Qwen-Max/Plus/Turbo, GLM-4-Plus, MiniMax-Text-01, and 30+ more models.

**Q: How is pricing calculated?**
A: Billed per token usage (priced per million tokens), with different rates per model. Check real-time pricing in the Model Market. Some models are priced lower than direct provider access through multi-channel aggregation.

**Q: Is streaming supported?**
A: Yes, fully supported. Set stream: true to receive Server-Sent Events (SSE) streaming responses, fully compatible with OpenAI's streaming format.

**Q: How is data security ensured?**
A: The platform only forwards requests — no conversation data is stored. API Keys are stored as SHA-256 hashes, upstream provider keys are AES-256 encrypted. Additional protections include IP blacklisting, rate limiting, TOTP 2FA, and login lockout.

**Q: What happens when an upstream provider goes down?**
A: Built-in circuit breaker detects 5 consecutive failures and opens for 30 seconds, automatically routing requests to backup channels. Non-streaming requests also get one automatic retry.

**Q: What is the Agent/Reseller program?**
A: Agents purchase tokens at wholesale prices and resell to downstream businesses at their own markup. The platform provides full customer management, analytics dashboards, redeem code issuance, and revenue analytics tools.

**Q: Can I use one API Key for all models?**
A: Yes. A single API Key can access all models your account has permission to use — just change the model parameter in your request.

**Q: What payment methods are supported?**
A: Wallet top-up, redeem codes (vouchers), and Alipay online payment are supported. Agents can also directly recharge for their downstream companies.

## Contact

For agent partnerships and business inquiries, visit https://token.bitbeam.cn/apply-agent.

## Technical Details

- Built with: Nuxt 3, Vue 3, TypeScript, Prisma ORM, ClickHouse, Redis
- API Format: OpenAI-compatible REST API with streaming support (SSE)
- Authentication: Bearer token (API Key) for API, JWT for dashboard
- Billing Model: Pre-deduct → Relay → Post-settle (industry standard pattern)
- Data Isolation: Company-level multi-tenancy with role-based access control