ModelRefs / Tree-of-Thoughts — Prompt Pattern
Tree-of-Thoughts — Prompt Pattern
Explore multiple reasoning branches and prune with a value function. Generate K candidate next-steps at each node, score them, and expand only the top-M.
Overview
Generate K candidate next-steps at each node, score them, and expand only the top-M. Useful for search, planning, and puzzles where greedy CoT fails.
When to use it: Plain CoT fails because the model commits to a wrong path early.
Pattern details
| Pattern class | reasoning |
|---|---|
| Difficulty | advanced |
| Also known as | tot, branching reasoning |
| Last reviewed | 2026-06-07 |
Known failure modes
- Score collapse — Self-ratings cluster near 1.0. Mitigation: Normalize scores or use pairwise comparisons.
- Compute explosion — Cost scales as k^depth. Mitigation: Cap depth and use aggressive pruning.
When not to use it
- Using ToT on tasks plain CoT solves.
Continue your research
Use these connected ModelRefs sections to compare alternatives, inspect implementation paths, and review the evidence and governance boundaries relevant to Tree-of-Thoughts — Prompt Pattern.
Frequently asked questions
When should I use the Tree-of-Thoughts prompt pattern?
Plain CoT fails because the model commits to a wrong path early.
What are common failure modes of Tree-of-Thoughts?
Score collapse • Compute explosion
Does Tree-of-Thoughts work across models?
Yes when the anatomy is preserved — see model compatibility matrix on the pattern page.