Agent Orchestration Stress Testing with Model Intelligence

The real test of agent orchestration is can you have a very intelligent model (i.e. Opus-4.6) hand-hold a very stupid model (i.e. Llama-3) into producing correct results. (Through perfect task decomposition, hyper-specific prompting, and relentless verification/retry loops). You'd never do this in production, obviously use a much smarter 'dumb' model where cost or tuning is a factor. But IMO this is an honest stress test; if your orchestration only works when the underlying model is smart enough to save you then you don't have good orchestration. Strip away model intelligence and see whether your system or your model is load-bearing.