STILL 3 TOOL 32B

🚀 STILL-3-TOOL-32B

STILL-3-TOOL-32B is a model that leverages Python code to aid the reasoning process. It achieves an accuracy of 81.70% on AIME 2024, matching the performance of o3 - mini and outperforming o1 and DeepSeek - R1. The code, model, and data of this project are all open - sourced.

🚀 Quick Start

We propose STILL-3-Tool-32B, which uses Python code to assist the reasoning process. During evaluation, it shows excellent performance on AIME 2024.

We open - source our code, model, and data.

For more details, please refer to our Notion page.

📚 Documentation

During evaluation, STILL-3-Tool-32B achieves 81.70% accuracy on AIME 2024, matching the performance of o3 - mini and outperforming o1 and DeepSeek - R1.

📄 License

Citation

Please kindly cite our report if they are helpful for your research.

@article{Slow_Thinking_with_LLMs_3_Tool,
  title={Tool Manipulation Significantly Enhances the Reasoning Ability of O1- and R1-like LLMs},
  author={RUCAIBox STILL Team},
  url={https://github.com/RUCAIBox/Slow_Thinking_with_LLMs},
  year={2025}
}