Tree-of-thought (ToT)
Tree-of-thought prompting extends chain-of-thought by exploring multiple reasoning paths in parallel, evaluating each, and pursuing the most promising — analogous to tree search in game-playing AI. ToT trades inference cost for accuracy on hard reasoning tasks where the linear reasoning of CoT can lock into wrong paths early.
ToT was introduced by Yao et al. (2023) and is most useful for problems with a large search space and clear evaluation per partial solution: puzzles, planning, theorem proving. The implementation involves multiple sampling, evaluator prompts that score each path, and a search strategy (BFS, DFS, beam) that prioritises exploration. The cost is significant — many more inference calls per query — so production ToT is reserved for high-value tasks where accuracy gain justifies the cost multiplier. Modern reasoning models incorporate ToT-like exploration internally without requiring explicit prompt engineering.
Related terms
- Chain-of-thought (CoT)
Chain-of-thought prompting asks the LLM to reason step by step before producing the final answer — 'let's think through this carefully' or 'show your work'.
- Autonomous agent
An autonomous agent is an AI system that pursues a goal over multiple steps with minimal human intervention — deciding what actions to take, executing them via tools, observing results, and iterating until the goal is achieved or judged unreachable.
- Large language model (LLM)
A large language model is a neural network trained on enormous text corpora to predict the next token given preceding tokens — typically with billions to trillions of parameters.