Template · Multi-model reasoning
Consensus Draft
Three models answer your question independently, then a synthesizer combines the strongest points into one response.
When to use it
Best for open-ended questions, research synthesis, and planning prompts where different models contribute complementary strengths.
The research behind it
This template is based on Mixture-of-Agents (Wang et al., 2024).
Related templates
Rank & Fuse
Three models generate candidate answers, a ranker scores them, then a fuser combines the strongest elements guided by the ranking.
FreeSmart Escalation
A fast model answers first. If the answer is uncertain or incomplete, stronger models are consulted and their answers are synthesized.
FreeBest-of-N
Five high-temperature responses are generated, scored, and the best one is selected.
Run Consensus Draft
Add your prompt and let the template do the rest. Free to start, no card.
Open in LLMWeave