Pick the proposer models that answer in parallel, then the aggregator that fuses them. Saved as data — usable instantly as a model name everywhere.
Same prompt, many models/presets at once. See who wins on quality and speed.
Every model your configured providers expose right now.