MAS-Orchestra: Understanding and Improving Multi-Agent Reasoning Through Holistic Orchestration and Controlled Benchmarks

🎉 News • 🔗 Links • 📝 Conceptual Overview • 📊 Results

✨ Getting Started • 🏋️ MAS-Orchestra •

🎈 Citation • 🌻 Acknowledgement • 📧 Contact

• [2025/05/06] We present the MAS-Orchestra [Project Page | Paper | Code]

🔗 Links

---

• 🏠 [Project Page]
• 📜 [Paper]
• 💻 [Code]

✨ Getting Started

---

🎄 Environment Setup

conda create -n mas-orchestra python==3.10
conda activate mas-orchestra

apt update && apt install -y wget curl

cd ./verl
./install.sh
pip install --no-deps -e .
pip install ray==2.49.2 --force-reinstall
pip install protobuf==4.25.8 --force-reinstall
pip install together
pip install math-verify[antlr4_13_2]
pip install antlr4-python3-runtime==4.9.3

pip install langchain-core langchain-together langchain-community duckduckgo-search tavily-python pydantic ddgs langchain_brightdata bs4
pip install pyserini faiss-gpu
pip install git+https://github.com/texttron/tevatron.git

🏋️ MAS-Orchestra

---

♟️ Example Training Script

export OPENAI_API_KEY={YourKey}
export TOGETHER_API_KEY={YourKey}
export WANDB_API_KEY={YourKey}
LOG_FILE={YourLogFile}

python -u -m mas_r1_reasoner.main_mas_r1 \
    --config-path=configs \
    --config-name=grpo_trainer \
    data.max_prompt_length=15000 \
    data.max_validation_prompt_length=15000 \
    data.val_files=data/browse_comp/test_subset_200.parquet \
    data.train_files=data/browse_comp/train_subset_1066.parquet \
    azr.mas_r1.use_llm_judge=True \
    data.raw_data=True \
    data.train_batch_size=64 \
    actor_rollout_ref.rollout.n=32 \
    azr.mas_r1.execution_success_weight=0.0 \
    azr.mas_r1.final_answer_weight=1.0 \
    azr.mas_r1.agent.model_name=gpt-oss-120b\
    azr.mas_r1.multiply_processes=0 \
    azr.mas_r1.max_ray_workers=1 \
    azr.problem_type=harmony_medium \
    azr.mas_r1.agent.init_archive=['COT','COT_SC','Reflexion','LLM_debate','WebSearch'] \
    trainer.val_before_train=True \
    trainer.test_freq=5 \
    trainer.save_freq=10 \
    actor_rollout_ref.model.path=Qwen/Qwen2.5-7B-Instruct \
    trainer.experiment_name=harmony_medium_grpo_7b_gpt_oss_120b_browse_comp_plus \
    $@ 2>&1 | tee -a "$LOG_FILE"

🎈 Citation

---

If you find MAS-Orchestra helpful, please consider starring this repo and citing our work. We would be very grateful!

@misc{Ke2026MASOrchestra,
        title        = {MAS-Orchestra: Understanding and Improving Multi-Agent Reasoning Through Holistic Orchestration and Controlled Benchmarks},
        author       = {Zixuan Ke and Yifei Ming and Austin Xu and Ryan Chin and Xuan-Phi Nguyen and Prathyusha Jwalapuram and Semih Yavuz and Caiming Xiong and Shafiq Joty},
        year         = {2026},
        eprint       = {2601.14652},
        archivePrefix= {arXiv},
        primaryClass = {cs.AI},
        note         = {Preprint; Work in Progress},
      }

🌻 Acknowledgement

---

This project received help from many researchers at Salesforce AI Research. We also thank thanks to the authors of the verl for their excellent contributions to the community!

📧 Contact

---

Feel free to contact Zixuan Ke via email: zixuan.ke@salesforce.com