kyegomez/tree-of-thoughts

What is it?

Key features

• 01Breadth-first and depth-first search over reasoning steps, with backtracking and pruning.
• 02Value and vote evaluation of candidate steps.
• 03Configurable branching factor, depth, beam width and score threshold.
• 04Works with any LiteLLM model that supports function calling.
• 05Returns the full search tree as JSON-serializable data.

Use cases

• Solving multi-step arithmetic or puzzle problems where a single model pass often fails.
• Exploring planning or scheduling options by keeping several partial solutions alive.
• Running case analysis that needs backtracking when a branch contradicts itself.
• Evaluating reasoning quality on math or logic benchmarks with checkable answers.

Quick start

1. 1

   Install the swarms package

   pip3 install -U swarms
2. 2

   Add your API key to a .env file in your working directory

3. 3

   Run the Quickstart Python code using TreeOfThoughts

   python examples/physics/hohmann_transfer.py

Things to know

• !Requires an API key from a model provider that supports function calling.
• !Multi-branch search makes many model calls, so cost grows quickly with settings like breadth and depth.
• !Tree of Thoughts now ships inside the larger Swarms framework, which adds a sizable dependency.

In collections

Topics

FAQ

What is tree-of-thoughts?

A Python implementation of Tree of Thoughts, which lets LLMs solve problems by searching over candidate reasoning steps instead of answering in one pass. Tree of Thoughts makes a language model reason by search. It proposes several candidate next steps, scores each one, prunes weak branches, and backtracks when a line of reasoning fails. This repository ships the TreeOfThoughts class, which now lives in the Swarms framework, and runs breadth-first or depth-first search over partial solutions.

Is tree-of-thoughts free to use? What license does it have?

tree-of-thoughts is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.

Is tree-of-thoughts still maintained?

Updated within 6 months. The last commit to tree-of-thoughts was on Sep 29, 2026.

What can I use tree-of-thoughts for?

Solving multi-step arithmetic or puzzle problems where a single model pass often fails. Exploring planning or scheduling options by keeping several partial solutions alive. Running case analysis that needs backtracking when a branch contradicts itself. Evaluating reasoning quality on math or logic benchmarks with checkable answers.

How do I install tree-of-thoughts?

Install the swarms package: pip3 install -U swarms Add your API key to a .env file in your working directory Run the Quickstart Python code using TreeOfThoughts: python examples/physics/hohmann_transfer.py

Where can I get tree-of-thoughts?

The source code and installation instructions are on GitHub: https://github.com/kyegomez/tree-of-thoughts