Skip to main content

Tutorials / Build a consensus draft that merges multiple models

Tutorial

Build a consensus draft that merges multiple models

Beginner5 min

Instead of picking one model’s answer, merge several into a single stronger draft. This is the pattern most people want when they say "use multiple models".

Before you start

  • An LLMWeave account
  • Comfort starting a weave from a template

Choose the consensus template

Start a New Weave and pick the consensus draft template. It's preconfigured to send your prompt to several models and then synthesize their outputs.

Set your prompt and models

Write the task and confirm the model set. More diverse models tend to produce a richer synthesis, since they bring different strengths.

Run and review the merged answer

The weave produces one synthesized draft that leads with the answer and folds in the strongest points from each model, without source-by-source attribution or repeated points.

Refine by chaining

Not quite right? Continue the conversation to push the draft further. The synthesis carries forward as the new starting point.

Keep going

Try it yourself

Put this pattern to work on your own task.

Create a weave