The architectural pattern of coordinating multiple AI models to process a single task, often involving parallel execution, routing, or ensemble methods to optimize performance, cost, or reliability.
Overview
- OpenRouter released a new Fusion API that fans out prompts to a panel of models in parallel and uses a judge model to select the best response, claiming to achieve ‘fabled level intelligence at half the price.’ (Source: AI Daily Brief host (reporting on OpenRouter), via AI Daily Brief, 2026-08-24)