MODEL ROUTE Keywords
Route LLM requests to different models based on task type, cost, or capability requirements.
Keywords
| Keyword | Purpose |
|---|---|
USE MODEL | Select a specific model for next request |
SET MODEL ROUTING | Configure routing strategy |
GET CURRENT MODEL | Get active model name |
LIST MODELS | List available models |
USE MODEL
USE MODEL "fast"
response = ASK "Quick question about the weather"
USE MODEL "quality"
response = ASK "Analyze this complex legal document"
SET MODEL ROUTING
SET MODEL ROUTING "auto"
SET MODEL ROUTING "cost"
SET MODEL ROUTING "manual"
Routing Strategies
| Strategy | Description |
|---|---|
manual | Explicitly specify model per request |
auto | Auto-select based on task complexity |
cost | Prefer cheaper models when possible |
quality | Always use highest quality model |
GET CURRENT MODEL
model = GET CURRENT MODEL
TALK "Currently using: " + model
LIST MODELS
models = LIST MODELS
FOR EACH m IN models
TALK m.name + " - " + m.description
NEXT
Configuration
Add to config.csv:
llm-models,default;fast;quality;code
model-routing-strategy,auto
model-default,claude-sonnet-4.5
model-fast,gemini-flash
model-quality,claude-opus-4.5
model-code,claude-sonnet-4.5
Example: Task-Based Routing
USE MODEL "code"
code_review = ASK "Review this function for bugs: " + code
USE MODEL "fast"
TALK "Here's what I found:"
TALK code_review