Orchestrator
TensorLang routing rules — control which model handles which request.
Routing Rules
1
Legal → Opus
skill == "Senior MA Attorney"claude-opus-4-5
2
Code → GPT-4o
task_type == "code_review"gpt-4o
3
Fast queries → Flash
token_estimate < 500gemini-2.0-flash
4
Cost cap fallback
monthly_spend > 80%claude-haiku-4-5
5
Default
trueclaude-sonnet-4-5
TensorLang Reference
# Route high-stakes legal queries to Opus when skill == "Senior MA Attorney" and token_estimate > 2000: model: claude-opus-4-5 temperature: 0.2 inject_skills: true