InternAgent-1.5: A Unified Agentic Framework for Long-Horizon Autonomous Scientific Discovery

Autonomous Discovery Across All Sciences

• Papers: InternAgent 1.0 | InternAgent 1.5
• Links: Website | HuggingFace

🔥 News

• 2026.2.12: 🔥 🔥 Leveraging the general capabilities of InternAgent 1.5, anyone can now submit their algorithm tasks for optimization by opening an issue/PR in this repository. We will regularly update the algorithm design results. For other scientific discovery tasks, please visit Intern-Discovery.

• 2026.2.10: 🔥 Official release of the InternAgent 1.5 Technical Report. InternAgent 1.5 achieves leading performance on scientific reasoning benchmarks including GAIA, HLE, GPQA, and FrontierScience, and supports end-to-end autonomous scientific discovery tasks across Physical, Biology, Earth, and Life Science domains, enabling both algorithm discovery and empirical discovery (dry/wet-lab experiments).

• 2025.10.13: InternAgent-1.0 code has been fully open-sourced, supporting end-to-end automation and autonomous evolution across 12 scientific research tasks.

• 2025.07.17: The source code of InternAgent has been partially open-sourced. The complete version of InternAgent (covering 12 types of tasks for autonomous scientific research) will be open-sourced soon. This code repository can be used for full-cycle autonomous scientific research, ranging from hypothesis generation to automated experimental execution.

• 2025.07.10: NovelSeek has been renamed to InternAgent. This change embodies our hopeful vision for autonomous scientific research framework, and we hope it will empower all researchers to achieve great scientific discoveries.

---

📖 Overview

InternAgent 1.5 is a unified autonomous system for end-to-end scientific discovery across both Algorithm Discovery and Empirical Discovery. Building on InternAgent 1.0, it organizes scientific inquiry into three coordinated subsystems: Generation (hypothesis construction via deep research), Verification (methodological evaluation via solution refinement), and Evolution (evidence-driven refinement via long-horizon memory). InternAgent 1.5 achieves leading performance on scientific reasoning benchmarks (GAIA, HLE, GPQA, FrontierScience, SGI-bench) and demonstrates sustained autonomous optimization across extended discovery cycles. The system supports algorithm discovery (agent memory, reinforcement learning, test-time scaling, ...) and empirical discovery workflows (dry-lab simulations and wet-lab experimentation) across Physical, Biological, Earth, and Life Sciences.

---

🌟 Core Features

InternAgent 1.5 is built on three foundational subsystems that enable autonomous scientific discovery:

🔍 Generation: Deep Research for Hypothesis Construction

• Autonomous literature analysis and knowledge synthesis across scientific domains
• Multi-source information integration from papers, code repositories, and domain-specific databases
• Structured hypothesis formulation grounded in existing scientific evidence

✅ Verification: Solution Refinement for Methodological Evaluation

• Systematic transformation of hypotheses into executable experimental protocols
• Automated code generation, debugging, and execution across computational and experimental environments
• Exception-guided intelligent error correction and iterative solution optimization

🔄 Evolution: Long-Horizon Memory for Evidence-Driven Refinement

• Persistent memory architecture that accumulates knowledge across extended research cycles
• Cross-iteration learning from experimental outcomes and methodological feedback
• Adaptive optimization that continuously refines hypotheses and experimental designs

🧩 Three-Subsystem Coordination

• Generation → Verification → Evolution forms a complete discovery cycle
• Seamless integration of dry-lab (computational modeling) and wet-lab (physical experimentation) workflows
• Extensible architecture supporting diverse tasks across Algorithm Discovery and Empirical Discovery

InternAgent 1.5 delivers end-to-end autonomous scientific discovery, enabling researchers to complete the full cycle—from hypothesis generation to experimental validation—across Physical, Biological, Earth, and Life Sciences.

---

🔬 Supported Research Tasks

Scientific Algorithm Discovery

• Suzuki–Miyaura Reaction Yield Prediction
• Transcription Prediction for Perturbation Response
• Power Flow Estimation
• Time Series Forecasting
• Molecular Dynamics Simulation
• Enhancer Activity Prediction

AI Algorithm Discovery

• Test-Time Scaling for LLM Reasoning
• Long-Term Memory Management for Agents
• Self-Distillation for Mathematical Reasoning
• Test-Time Reinforcement Learning

Empirical Discovery

• Automated Climate Diagnostics
• Climate Downscaling Optimization
• Biological Evidence Synthesis for Target Discovery
• Hypothesis Generation and Target Prioritization
• Fluorescent Protein Engineering
• Automated Reaction Outcome Prediction
• Generative Scaffold Hopping And more...

---

🎉 Benchmark Results

Results on Al Research Tasks

InternAgent consistently improves upon the baseline and outperforms Dolphin across all tasks, spanning AI and scientific domains.

Max Performance

Task	Metric	Baseline	Dolphin	InternAgent
AutoRYP	R² ↑	27.6	31.8 (+4.2)	35.4 (+7.8)
AutoMD	Forces-MAE ↓	0.158	0.152	0.148
AutoPower	RMSE ↓	0.00473	0.00455	0.00426
AutoTSF	MAE ↓	0.4382	0.4627	0.4331
AutoTPPR	MSE ↓	0.197	0.173	0.146
AutoEAP	HK-PCC ↑	0.65	0.76	0.79
AutoSenCls	Acc ↑	91.0	92.5 (+1.5)	93.5 (+2.5)
Auto2DCls	Top-1 Acc ↑	81.2	82.0 (+0.8)	83.3 (+2.1)
Auto3DCls	OA ↑	91.0	93.9 (+2.9)	95.5 (+4.5)
Auto2DSeg	mIoU ↑	78.8	-	81.0 (+2.2)
AutoPCDet	mAP ↑	65.0	-	65.9 (+0.9)
AutoVLM	QA ↑	67.1	-	67.6 (+0.5)

Average Performance

Task	Metric	Baseline	Dolphin	InternAgent
AutoRYP	R² ↑	27.6	31.3 (+3.7)	33.5 (+5.9)
AutoMD	Forces-MAE ↓	0.158	0.155	0.152
AutoPower	RMSE ↓	0.00473	0.00459	0.00447
AutoTSF	MAE ↓	0.4382	-	0.4346
AutoTPPR	MSE ↓	0.197	0.179	0.170
AutoEAP	HK-PCC ↑	0.65	0.73	0.77
AutoSenCls	Acc ↑	91.0	91.8 (+0.8)	92.5 (+1.5)
Auto2DCls	Top-1 Acc ↑	81.2	81.8 (+0.6)	82.2 (+1.0)
Auto3DCls	OA ↑	91.0	92.0 (+1.0)	93.4 (+2.4)
Auto2DSeg	mIoU ↑	78.8	-	80.1 (+1.3)
AutoPCDet	mAP ↑	65.0	-	65.7 (+0.7)
AutoVLM	QA ↑	67.1	-	67.6 (+0.5)

---

🧪 GAIA, GPQA-Diamond, FrontierScience and HLE Benchmarks

InternAgent-1.5 achieved state-of-the-art results across multiple benchmarks.

Humanity's Last Exam (HLE)

Setting	Model	Math	Bio/Med	CS/AI	Physics	Human.	Chem.	Engineer.	Other	Avg.
Text-Only	Deepseek-R1	9.30	8.60	7.40	5.80	11.00	5.60	10.30	7.50	8.60
	Gemini-3-pro-preview	45.08	26.13	26.79	32.67	44.04	34.65	29.69	32.39	38.00
	InternAgent-1.5	48.96	30.63	29.46	34.16	44.56	30.69	28.13	37.50	40.87
All-Set	o4-mini	19.00	11.40	12.90	12.60	9.10	12.70	12.60	6.90	14.30
	GPT-5	31.00	22.10	24.90	21.70	20.60	16.40	14.40	18.00	24.80
	Gemini-3-pro-preview	44.76	27.14	29.05	31.30	42.92	40.00	32.43	34.33	38.04
	InternAgent-1.5	48.09	30.36	30.71	33.04	42.47	34.55	30.63	38.63	40.00

---

FrontierScience Benchmark

Method	Olympiad (avg N=20)				Research (avg N=30)
	Bio	Chem	Phy	All	Bio	Chem	Phy	All
o4-mini	47.00±14.90	65.00±6.40	53.40±4.50	57.40±3.30	9.67±5.47	8.17±4.37	0.83±2.27	6.20±2.54
InternS1-235B	17.00±12.69	52.88±4.05	50.40±3.88	48.05±2.84	4.50±4.35	11.00±3.74	2.67±3.35	6.06±2.30
Mirothinker-v1.5-30B-A3B	22.86±4.52	69.64±7.49	54.86±3.18	57.57±3.66	8.17±6.39	8.50±6.21	5.83±4.10	7.50±3.77
DeepSeek-V3.2-Thinking	26.50±7.26	72.25±3.25	66.30±2.63	64.70±2.41	2.50±3.10	16.33±4.64	1.40±2.70	6.84±1.88
Qwen3-235B-A22B-Thinking	24.00±9.17	61.13±6.05	57.10±4.79	55.40±3.68	10.17±5.08	10.00±6.32	1.58±2.41	7.34±3.37
Qwen3-30B-A3B-Thinking	13.50±9.10	47.25±4.47	42.70±3.65	41.60±2.94	1.50±2.93	2.00±3.32	0.70±1.79	1.41±1.52
InternAgent-1.5	46.00±8.00	85.50±3.67	76.80±2.99	77.20±3.06	10.33±4.64	22.00±6.00	3.67±2.87	12.00±2.49

---

GPQA-Diamond Benchmark

Agent	Bio	Chem	Phys	Avg.
Base Models
Qwen-3-8B	-	-	-	44.44
Qwen3-32B	-	-	-	49.49
Qwen3-235B	-	-	-	47.47
Intern-S1	89.47	59.49	93.02	78.26
Deepseek-R1	63.16	76.34	91.86	82.32
o4-mini	78.95	63.44	94.19	78.28
GPT-5	84.21	76.34	95.35	85.35
React Model with Tools
WebShaper	47.37	52.69	81.40	64.65
MiroThinker	84.21	75.27	95.35	84.85
Tongyi DR	78.95	67.74	95.35	80.30
InternAgent-1.5	84.21	79.57	96.51	87.37

---

🚀🚀 Getting Started with InternAgent 1.5

Algorithm Discovery Tasks

For algorithm discovery tasks such as Reinforcement Learning, Test-time Scaling, Agent Memory... we currently support access to InternAgent 1.5 by submitting an issue or pull request in this repository. Please describe your optimization task, and we will regularly update the algorithm design results.

Empirical Discovery Tasks

For empirical discovery tasks including computational modeling, dry-lab simulations, and wet-lab experimentation across Physical, Biological, Earth, and Life Sciences, please visit Intern-Discovery.

Stay tuned for more updates as we expand access and capabilities!

🚀 Getting Started with InternAgent 1.0

Installation

conda create -n InternAgent python=3.11
conda activate InternAgent

# Install PyPI requirements
pip install -r requirements.txt

# Install aider
python -m pip install -U --upgrade-strategy only-if-needed aider-chat

Set Your API Key

Rename .env.example to .env and fill in your API keys:

mv .env.example .env

Start Your Research Project

./scripts/run_pipeline.sh

Configuration Tips:

• Modify configs/config.yaml to customize your research project
• Results will be saved in the results/ directory
• Check logs in the logs/ directory
• To skip idea generation, refer to scripts/run_skip-idea.sh
• Visualize idea evolution using internagent/vis_tree.py

About Research Tasks

We provide the tasks mentioned in our technical report as examples. Each task has different training environments and datasets. Please refer to the code in each task's folder for configuration details.

---

📝 Citation

@article{feng2026internagent,
  title={InternAgent-1.5: A Unified Agentic Framework for Long-Horizon Autonomous Scientific Discovery},
  author={Shiyang Feng and Runmin Ma and Xiangchao Yan and Yue Fan and Yusong Hu and Songtao Huang and Shuaiyu Zhang and Zongsheng Cao and Tianshuo Peng and Jiakang Yuan and Zijie Guo and Zhijie Zhong and Shangheng Du and Weida Wang and Jinxin Shi and Yuhao Zhou and Xiaohan He and Zhiyin Yu and Fangchen Yu and Bihao Zhan and Qihao Zheng and Jiamin Wu and Mianxin Liu and Chi Zhang and Shaowei Hou and Shuya Li and Yankai Jiang and Wenjie Lou and Lilong Wang and Zifu Wang and Jiong Wang and Wanghan Xu and Yue Deng and Dongrui Liu and Yiheng Wang and Wenlong Zhang and Fenghua Ling and Shufei Zhang and Xiaosong Wang and Shuangjia Zheng and Xun Huang and Siqi Sun and Shuyue Hu and Peng Ye and Chunfeng Song and Bin Wang and Conghui He and Yihao Liu and Xin Li and Qibin Hou and Tao Chen and Xiangyu Yue and Bin Wang and Liang He and Dahua Lin and Bowen Zhou and Bo Zhang and Lei Bai},
  journal={arXiv preprint arXiv:2602.08990},
  year={2026}
}

@article{team2025internagent,
  title={InternAgent: When Agent Becomes the Scientist--Building Closed-Loop System from Hypothesis to Verification},
  author={Team, InternAgent and Zhang, Bo and Feng, Shiyang and Yan, Xiangchao and Yuan, Jiakang and Ma, Runmin and Hu, Yusong and Yu, Zhiyin and He, Xiaohan and Huang, Songtao and others},
  journal={arXiv e-prints},
  pages={arXiv--2505},
  year={2025}
}

@article{hu2025flowsearch,
  title={FlowSearch: Advancing deep research with dynamic structured knowledge flow},
  author={Yusong Hu and Runmin Ma and Yue Fan and Jinxin Shi and Zongsheng Cao and Yuhao Zhou and Jiakang Yuan and Xiangchao Yan and Wenlong Zhang and Lei Bai and Bo Zhang},
  journal={arXiv preprint arXiv:2510.08521},
  year={2025}
}

@article{du2025automlgen,
  title={AutoMLGen: Navigating Fine-Grained Optimization for Coding Agents},
  author={Shangheng Du and Xiangchao Yan and Dengyang Jiang and Jiakang Yuan and Yusong Hu and Xin Li and Liang He and Bo Zhang and Lei Bai},
  journal={arXiv preprint arXiv:2510.08521},
  year={2025}
}