OpenResearcher/OpenResearcher-30B-A3B
OpenResearcher/OpenResearcher-30B-A3B is a 30 billion parameter agentic large language model developed by OpenResearcher, fine-tuned from NVIDIA-Nemotron-3-Nano-30B-A3B-Base-BF16. It is specifically designed for long-horizon deep research tasks, leveraging a 32768 token context length. The model excels at complex research trajectories, achieving 54.8% accuracy on BrowseComp-Plus, outperforming models like GPT-4.1 and Claude-Opus-4.
Loading preview...
OpenResearcher-30B-A3B: Agentic Model for Deep Research
OpenResearcher-30B-A3B is a 30 billion parameter agentic large language model developed by OpenResearcher, specifically fine-tuned for long-horizon deep research tasks. It is based on the NVIDIA-Nemotron-3-Nano-30B-A3B-Base-BF16 architecture and utilizes a substantial 32768 token context length.
Key Capabilities & Differentiators
- Specialized for Deep Research: This model is explicitly designed and fine-tuned for complex, multi-step research trajectories, distinguishing it from general-purpose LLMs.
- High Performance on Research Benchmarks: It achieves an impressive 54.8% accuracy on the BrowseComp-Plus benchmark, surpassing several prominent models including GPT-4.1, Claude-Opus-4, Gemini-2.5-Pro, DeepSeek-R1, and Tongyi-DeepResearch.
- Extensive Fine-tuning: The model was fine-tuned on 96K entries from the OpenResearcher dataset, which consists of over 100-turn research dialogues distilled from GPT-OSS-120B using native browser tools.
- Adopted by NVIDIA: The OpenResearcher framework has been adopted by NVIDIA's Nemotron family of models, indicating its robust design and potential.
Ideal Use Cases
- Automated scientific literature review and synthesis.
- Complex information retrieval and summarization requiring multi-step reasoning.
- Developing AI agents for research-intensive applications.
- Tasks requiring long-horizon planning and execution in research domains.