1/ Break every incoming prompt into 3 complexity tiers:
• 80% Tier 1 (Intent classification, JSON extraction, summarization): Route to Gemini 2.0 Flash ($0.10/$0.40) or Llama 3.3 70B ($0.12/$0.30)
• 15% Tier 2 (Complex code & multi-step tool use): Route to Claude 3.7 Sonnet ($3/$15 w/ $0.30 ca...
Replies1 Reposts0 Likes0