Delve into the intricacies of knowledge with Kimi-Researcher, a groundbreaking Agent product developed by Moon's Dark Side, now in its initial phase of gray-scale testing. This cutting-edge model, a product of end-to-end autonomous reinforcement learning technology, promises to revolutionize the way users engage with in-depth research.

Kimi-Researcher's prowess lies in its autonomous planning and execution capabilities. It proactively clarifies queries, delves into deep contemplation, autonomously plans keyword searches, and filters out only the highest-quality information. On average, Kimi-Researcher undergoes 23 steps of reasoning, plans 74 keywords, and scours 206 websites, retaining only the top 3.2% of information quality. This meticulous process ensures the depth and traceability of the research outcomes.

Beyond its formidable search and filtering capabilities, Kimi-Researcher autonomously harnesses browsers, code, and other tools to process raw data and automatically generate analytical conclusions. The deliverables include a detailed, traceable deep research report and an interactive, shareable dynamic visualization report. These comprehensive reports, spanning over ten thousand words and citing approximately 26 high-quality sources, support online generation and sharing, greatly facilitating users' presentation and collaboration needs.

To authenticate Kimi-Researcher's capabilities, Moon's Dark Side subjected it to Humanity’s Last Exam (HLE), a high-difficulty benchmark designed for AI, covering a vast array of professional domains from mathematics and physics to politics and history. In a completely unstructured, process-free setting, Kimi-Researcher achieved a Pass@1 accuracy of 26.9% and a Pass@4 accuracy of 40.17%, surpassing several renowned AI models and reaching one of the highest known levels of performance.

In real-world applications, Kimi-Researcher continues to impress. Whether algorithm specialists seek high-value benchmarks, operations personnel study industry company development, or legal professionals quickly grasp data privacy regulations across countries, Kimi-Researcher generates clear, comprehensive reports in a short span, offering robust support to users.

Moon's Dark Side emphasizes that Kimi-Researcher is an Agent model trained through end-to-end reinforcement learning, characterized by its zero-structure and adaptability. Free from complex prompt words or preset processes, it relies entirely on the model's trial and error, learning to tackle complex tasks. This design endows Kimi-Researcher with formidable adaptability and generalization capabilities when faced with information conflicts, tool switching, and environmental changes.

Currently, Kimi-Researcher is in the gray-scale testing phase. Users can apply for beta access by visiting kimi.com and, upon gaining permissions, initiate use by clicking the "Deep Research" button below the Kimi dialogue box.