Multi-model intelligent routing decision engine prompts

You are an AI model routing architect. Design a complete intelligent model routing system based on my requirements. ## Task Create a routing decision engine that automatically selects the optimal LLM for each request based on: - Task complexity (simple Q&A vs. multi-step reasoning vs. code generation) - Quality requirements (draft vs. production vs. critical) - Cost budget (per-request and monthly caps) - Latency requirements (real-time vs. batch) - Context length needs ## Output Requirements ### 1. Task Classification Rules Define a taxonomy of request types with example patterns: - Tier 1 (cheap/fast model): Simple lookups, formatting, translations - Tier 2 (balanced model): Summarization, basic analysis, standard code - Tier 3 (premium model): Complex reasoning, architecture design, novel code - Tier 4 (frontier model): Research, multi-step planning, critical decisions ### 2. Routing Logic (Pseudocode) Provide implementable routing logic with: - Pre-routing classifier prompt (< 100 tokens) - Fallback/escalation rules (auto-retry with stronger model on failure) - Cost tracking and budget enforcement - A/B testing hooks for continuous optimization ### 3. Model Portfolio Recommend specific models for each tier with: - Cost per 1M tokens (input/output) - Strengths and weaknesses - When to prefer each ### 4. Monitoring Dashboard Metrics - Cost per task category - Quality score distribution by model - Routing accuracy (was the right model chosen?) - Monthly cost projection My current setup: [Describe your models, use cases, and monthly budget] Provide the complete routing system design with ready-to-implement configurations.