Model Router for customer APIs

Multi-protocol AI model gateway

Use one API key to access global LLM resources through OpenAI-compatible, Anthropic-compatible, and DashScope-compatible endpoints. Manage accounts, models, routing, balance top-ups, recharge records, security policies, and observability from one tenant workspace.

Unified customer entry

Customers maintain one tenant account and API key, then manage model access, balance top-ups, and recharge records as needed.

/portal/signup /portal/login /portal/recharge
Model Router

Centralized authentication, usage metering, cost attribution, smart routing, fallback, rollout control, and provider model mapping.

routing billing observability
Multi-provider model resources

Operations teams manage providers, regions, protocols, credentials, customer-facing model names, and model availability while tenants call one gateway.

DashScope MiniMax OpenAI compatible
🔑
Unified API key
Use one key for authorized models. Tenant login uses email and password; API keys are only for programmatic calls.
🔁
Multi-protocol inference
OpenAI, Anthropic, and DashScope-compatible routes are mounted behind one platform, reducing migration cost.
💰
Unified billing
Manage account balance, recharge orders, recharge records, usage statistics, and billing vouchers in one place.
Smart routing
Route requests by model, provider, region, retry behavior, and fallback policy at runtime.
Tenant workspace
Customers enter real tenant workflows from the website LIVE

API key management

Create keys by model scope: automatic model matching, selected model sets, or a single fixed model.

Account balance and top-up

View account balance after login, create amount-based recharge orders, and review Recharge records and vouchers.

Usage and billing

Review requests, input tokens, output tokens, model-level cost, and billing exports.

Available models and access

Models come from provider resources configured by operations, with customer-facing display names in the tenant portal.

Multi-protocol inference endpoints API
Customers use the platform API key and choose the endpoint by protocol.
POST https://xbingo.xin/v1/chat/completions POST https://xbingo.xin/v1/messages POST https://xbingo.xin/api/v1/services/aigc/text-generation/generation
Platform capabilities
Model aggregation and provider configuration OPS

Operations manage provider regions, base URLs, protocols, credentials, model catalogs, customer-facing display names, and route bindings.

🏢Platform Admin Login
Security policy LIVE

Tenants configure content safety, access control, data protection, and runtime policies while programmatic calls still use the unified API.

🔐Tenant Security Policy
Observability LIVE

Aggregate API requests, models, cost, latency, errors, and routing decisions to isolate provider and model-path issues.

📊Tenant Observability
Commercial entry points
Registration and login LIVE

Tenants can self-register and log in with email and password. API keys are not used for human login.

Recharge entry LIVE

Top up account balance by amount, then review recharge records and vouchers. The public website does not expose package purchase flows.

💳Account Balance, Recharge Records, and Vouchers
Agent and provider onboarding OPS

Agent commercialization and provider configuration are managed in operations. Customers only see callable models.

🌐Agent Login
API status
GET /health/live
Liveness probe — service process health
GET /health/ready
Readiness probe — database connectivity status
GET /v1/models
Model list — requires API key authentication