Model synthesis is a new opt-in server-side tool in DigitalOcean Inference Engine. Run multiple models on the same inference request and get a single synthesized response, no custom orchestration required. Built for complex, high-reasoning tasks, it works with the existing Chat Completions, Responses, and Messages APIs. Choose an optimized preset or configure the panel and synthesizer directly.