Tree of Thoughts — deliberate problem solving with branching reasoning
Explore multiple reasoning paths simultaneously, evaluate each branch, and backtrack when a path leads to a dead end.
Reasoning4.1

Prompt template
Solve the problem by exploring multiple reasoning paths.
For each step:
1. Generate 3 possible next thoughts
2. Evaluate each thought (1-10) for progress toward the goal
3. Select the top 2 thoughts to continue
4. Backtrack if all paths score below 5
Problem: {problem}
Step 1 thoughts:How to use
- 1Define the problem and success criteria clearly
- 2Generate 3 candidate thoughts at each node
- 3Score each thought on relevance and progress
- 4Prune low-scoring branches, keep top 2
- 5Backtrack when no path scores above threshold
Pro tips
- Works best for puzzles, math, and strategic planning
- Use a separate evaluation prompt for objective scoring
- Keep branch width at 2-3 to manage token cost
- Record the full tree for post-hoc analysis
Tree of ThoughtsReasoningProblem Solvingchatgpt
Ratings & Reviews
Loading reviews…
Related Skills

Reasoning
Chain-of-Thought — let the model reason step by step
Ask the AI to think out loud and solve problems in steps to dramatically improve accuracy on reasoning tasks.
ReasoningProblem SolvingMath
4.8
Reasoning
Self-Consistency — sample multiple paths and vote on the answer
Generate several independent reasoning paths and select the answer that appears most frequently, reducing errors from single-path reasoning.
Self-ConsistencyEnsembleAccuracy
4.5
Agent
ReAct Prompting — reasoning + acting for tool-using agents
Alternate between Thought (reasoning) and Action (tool call) steps so the AI can solve multi-step tasks with external tools.
ReActAgentTool Use
4.6