Customize Model Combinations, Auto Failover
Create virtual model names, bind multiple real models, auto-schedule with active-standby/round-robin/random. auto-switches when a model fails
Create Your Smart Routes in 3 Steps
Create a route in Console > Smart Routes
Select a strategy (Active-Standby / Round-Robin / Random), add models, set priorities
Use {ID Code}/{Route Name} to call
model: abc123/k2w8xSystem schedules by strategy, auto-switches on failure
Three Scheduling Strategies
Choose the best scheduling method for your scenario
main_backup
Tries models by priority from high to low. Primary model is preferred; on failure, automatically switches to backup. Ideal for scenarios requiring high stability.
round_robin
Each request rotates to the next model, evenly distributing load. Ideal for spreading request volume and preventing single model overload.
random
Randomly selects a model for each request. Simple and efficient. Ideal for multiple models of equal quality where no specific priority is needed.
How It Works
Strategy Scheduling → Model Attempts → Failover → Return Result
Parse Route Name
Recognize the {ID Code}/{Route Name} format, look up the user's route config and its bound model list
Strategy Scheduling
Generate a model attempt queue based on strategy (Active-Standby / Round-Robin / Random), sorted by priority
Sequential Attempts
Try models in queue order; on failure, auto-switch to the next one. Supports channel-level failover
Return Result
Returns the first successful model result to the user. Fully transparent — same experience as calling a single model
Use Cases
When to Use Smart Routes?
Auto-switches to backup when primary model unstable, ensuring business continuity. for critical calls that cannot fail in production.
Multiple models of equal quality are used in rotation, spreading request volume to avoid triggering rate limits or quota exhaustion.
Prefer lightweight, cheaper models; fall back to more expensive but stable models on failure. Balances cost and quality.
Provide a fixed virtual model name for teams or apps, backend model changes don't affect frontend code. switch models by changing route config.
Auto Failover
Automatically switches to the next model on failure
Channel-Level Switch
Failover across multiple channels within the same model
Flexible Strategy
Choose from Active-Standby / Round-Robin / Random
Transparent
Same experience as calling a single model
Billing
Independent Billing per Model
Each model is billed according to its own route pricing. Plan users are deducted from their plan call count. Failed models are not charged.
Single Request, Single Log Entry
Regardless of how many models were tried, only one API log entry is created. The switching process is recorded in failover details.
FAQ
What's the difference between Smart Routes and Fusion?
Smart Routes tries models sequentially(returns first success), Fusion calls all models in parallel then judge fuses. Smart Routes prioritizes availability, Fusion prioritizes quality.
What is the ID Code in the call name?
The ID Code is your account's unique identifier (User.InviteCode). You can copy the full call name from the Console > Smart Routes page.
Does it support streaming output?
Fully supported. Both streaming and non-streaming work. The switching process is transparent to the client. Cherry Studio, Cursor, and other tools work seamlessly.
Can I select any model?
You can select from all platform routes, Free Route, and plan-included models. Go to console'Smart Routes'page to view available models.
Will switching affect speed?
Next model only tried when primary fails, No extra latency normally. First request establishes channel stickiness, subsequent requests prefer last successful channel.
Create your first smart route
Go to console'Smart Routes'page to configure custom model combinations
Create Now