Evaluate LLM tradeoffs across latency, cost, reasoning capabilities, and context window requirements.
