Yuyao Technology Launches "MatterSeek" Globally; Platform Now Available for Free to Researchers Worldwide

BEIJING, CHINA, October 28, 2026 /EINPresswire.com/ --

Available simultaneously on the company’s overseas official website and the Hugging Face AI cloud platform, MatterSeek brings AI-driven materials R&D from “year-scale” cycles to “month-scale” efficiency.

On October 29, MatterSeek—an AI-powered materials R&D platform developed by Yuyao Technology—officially launched globally via its international website and the Hugging Face is the world’s leading open-source AI model hosting and collaboration platform. The software is now available for free download, open to researchers and scientists across the globe. This milestone marks a critical step in Yuyao Technology’s global expansion within the field of foundational AI innovation for new materials, accelerating the transition of materials R&D from traditional, years-long development cycles to a timeframe measured in months.

Materials R&D has long been hindered by persistent, industry-wide challenges, including lengthy development cycles, fragmented data, low experimental efficiency, and the difficulty of making predictions across multiple scales. Under the traditional "trial-and-error" paradigm, bringing a new material from the laboratory to mass production typically takes 10 to 15 years. Furthermore, materials data is often scattered and heterogeneous, making systematic mining and reuse difficult; manual synthesis and characterization processes are inefficient, and high-throughput parallel experimentation remains challenging to implement. Additionally, there has long been a lack of effective, practical tools for predicting material properties across scales—from the atomic level to macroscopic performance.

To address these pain points, Yuyao Technology developed MatterSeek, an AI+MGI (Materials Genome Initiative) R&D platform. Powered by the company’s proprietary HNN (Hierarchical Neural Network), the platform spans three major R&D domains: inorganic materials, organic materials, and biopharmaceuticals. It stands as one of the few large-scale, non-language-based predictive models for new materials driven by physical reasoning. Unlike mainstream large language models, MatterSeek integrates the entire "prediction–synthesis–characterization" workflow, covering the full value chain of materials R&D. By leveraging physical models as a foundational engine—complemented by the company’s own laboratories to create a closed-loop system for real-world physical experimentation—MatterSeek offers a systematic solution to the four core challenges facing the materials industry, making it a unique and highly valuable asset in the global "AI for Science" landscape.

In terms of functionality, MatterSeek establishes a complete R&D closed loop—from intelligent input to final output—significantly lowering the barrier to entry for digital-driven materials research and development.

Yuyao Technology Launches "MatterSeek" Globally; Platform Now Available for Free to Researchers Worldwide
On October 29, MatterSeek—an AI-powered materials R&D platform developed by Yuyao Technology—officially launched globally via its international website and the HuggingFace AI cloud platform. The software is now available for free download, open to researchers and scientists across the globe. This milestone marks a critical step in Yuyao Technology’s global expansion within the field of foundational AI innovation for new materials, accelerating the transition of materials R&D from traditional, years-long development cycles to a timeframe measured in months.

Materials R&D has long been hindered by persistent, industry-wide challenges, including lengthy development cycles, fragmented data, low experimental efficiency, and the difficulty of making predictions across multiple scales. Under the traditional "trial-and-error" paradigm, bringing a new material from the laboratory to mass production typically takes 10 to 15 years. Furthermore, materials data is often scattered and heterogeneous, making systematic mining and reuse difficult; manual synthesis and characterization processes are inefficient, and high-throughput parallel experimentation remains challenging to implement. Additionally, there has long been a lack of effective, practical tools for predicting material properties across scales—from the atomic level to macroscopic performance.

To address these pain points, Yuyao Technology developed MatterSeek, an AI+MGI (Materials Genome Initiative) R&D platform. Powered by the company’s proprietary HNN (Hierarchical Neural Network), the platform spans three major R&D domains: inorganic materials, organic materials, and biopharmaceuticals. It stands as one of the few large-scale, non-language-based predictive models for new materials driven by physical reasoning. Unlike mainstream large language models, MatterSeek integrates the entire "prediction–synthesis–characterization" workflow, covering the full value chain of materials R&D. By leveraging physical models as a foundational engine—complemented by the company’s own laboratories to create a closed-loop system for real-world physical experimentation—MatterSeek offers a systematic solution to the four core challenges facing the materials industry, making it a unique and highly valuable asset in the global "AI for Science" landscape.

In terms of functionality, MatterSeek establishes a complete R&D closed loop—from intelligent input to final output—significantly lowering the barrier to entry for digital-driven materials research and development.

Development Scenarios Across the Materials Landscape

Because MatterSeek predicts from physics rather than language patterns, one model can be applied across a wide range of materials development scenarios — from atomic-scale composition screening to formulation-level optimization. Beyond the commercial deployments above, the platform is built to support R&D teams in the following directions:

Semiconductor & Microelectronics — screening low-k and high-k dielectrics, wide-bandgap semiconductors, and packaging or thermal-interface materials against dielectric constant, thermal stability, and process-compatibility targets.

Battery & Energy Storage — exploring cathode, anode, and solid-electrolyte compositions, and ranking candidates by ionic conductivity, electrochemical window, and thermal stability before committing to synthesis.

Photovoltaics & Solar — evaluating absorber layers, passivation films, and encapsulation materials, treating bandgap, defect tolerance, and long-term stability as joint constraints rather than sequential trial and error.

Catalysis & Chemicals — mapping catalyst and support compositions against activity, selectivity, and durability descriptors, and narrowing the experimental space before bench-scale testing.

Polymers & Organic Materials — linking monomer and backbone choices to glass-transition temperature, dielectric behavior, mechanical performance, and degradation profiles — including recyclable and biodegradable formulations.

Metals & Alloys — predicting phase stability, melting point, and high-temperature performance for superalloys, lightweight alloys, and structural metals, and flagging promising compositions for melting and casting trials.

Ceramics & Inorganic Materials — screening structural and functional ceramics — piezoelectric and dielectric compositions among them — for thermal shock resistance, sintering behavior, and electromechanical response.

Biologics & Pharma — supporting formulation and excipient selection, salt and polymorph screening, and stability assessment for biologic and small-molecule development, where the candidate space is vast and each experiment is costly.

Across all of these directions, users work in a single interface: describing the target in natural language, constraining the element space on the visualized periodic table, and comparing candidate compositions across multiple predicted properties — so that a development program can be sequenced in months rather than years.

At the underlying technology level, MatterSeek leverages the HNN algorithm to achieve few-shot prediction. It incorporates 1,035 built-in material descriptors, a self-developed model parameter rate of 92%, and 188 vertical-domain sub-models, reinforcing the platform’s technological foundation.
YuYao (Beijing) Technology Co., Ltd. is a national high-tech enterprise and a Beijing “specialized, refined, distinctive and innovative” (SRDI) small and medium-sized enterprise deeply engaged in China’s “AI + new materials” cross-innovation field. The company focuses on the intersection of artificial intelligence and new materials; its core businesses comprise new-materials large-model R&D, supporting technical services, and the industrialization of new-materials outcomes. The team has filed 27 national patents cumulatively. Its main products and services cover self-developed AI materials models, high-precision experimental databases for new materials, and customized new-materials R&D services, empowering intelligent materials R&D across the board.

Effective immediately, researchers worldwide can obtain free download access to the MatterSeek client, YuYao Technology’s flagship materials AI software, through the company’s overseas official website and the Hugging Face AI cloud platform. Looking ahead, YuYao Technology will continue to iterate and optimize its model capabilities in line with the real needs of global research users, working alongside researchers around the world to expand the boundaries of new-materials innovation.

Address: Room 912-5, 8th Floor, Building 3, District 4, No. 22, Jinghai 4th Road, Beijing Economic and Technological Development Zone, Beijing
Official Website: https://www.yuyaomatterseek.net/

Wang Yang
Yuyao(Beijing)Technology Co., Ltd.
matterseek@yuyaomatterseek.cn