On August 31, the Intelligent Chemical Industry Large Model 3.0 Pro was officially released, co-developed by the Dalian Institute of Chemical Physics (DICP) of the Chinese Academy of Sciences, iFLYTEK, Alibaba Cloud, and other partners. As the latest iteration of China's first large model for the chemical industry, 3.0 Pro marks a significant leap from a "professional knowledge Q&A model" to an "intelligent task execution system for chemical engineering," upgrading from a "professional assistant" to an "intelligent engineering partner" capable of collaborating with professionals on complex tasks.
Bridging the "Valley of Death": From Lab to Factory
The chemical industry is characterized by complex processes, large scale spans, diverse data types, and strong engineering constraints, making general-purpose models insufficient for professional needs. A catalytic reaction typically takes decades to move from the lab to the factory through pilot and industrial trials. Many promising lab results fail during the pilot stage—what the industry calls the "Valley of Death".
Since releasing version 1.0 in March 2024, DICP has continuously iterated the model. "If we can build a new paradigm of 'lab-virtual factory-physical factory' to digitally simulate the pilot stage first, we can save significant time and resources," said Ye Mao, a researcher at DICP and head of the intelligent chemical model team-3.
3.0 Pro's core breakthrough is its four-layer technical architecture: "Foundation Model — Agent — Professional Skills & Tools — Application Scenarios" :
The Foundation Model serves as the cognitive core, handling professional knowledge understanding, multimodal information parsing, solution generation, and task planning.
The Agent acts as the executor, breaking down tasks, invoking tools, verifying results, and making dynamic adjustments—forming a closed-loop capability spanning cognition, reasoning, planning, execution, and validation.
Performance Leap: 400+ Agents Validated by 14 Million API Calls
3.0 Pro integrates over 400 chemical agents and tools, trained on tens of billions of tokens of high-quality chemical Q&A data and hundreds of billions of tokens of specialized chemical corpus.
According to a multi-dimensional evaluation system for the chemical industry, 3.0 Pro achieved 81.96% accuracy in text Q&A and 80.75% in multimodal Q&A, representing relative improvements of 20.2% and 31.4% over version 3.0.
To date, over 300 chemical enterprises, universities, and research institutes have registered to use the model, with cumulative API calls exceeding 14 million.
DICP has also built China's first full-chain big data center covering "R&D-Design-Production-Market" for the petrochemical industry, and the country's first thousand-ton-level intelligent pilot platform in Dalian Changxing Island, capable of generating real-condition catalytic reaction and process data on demand.
Industry-Academia-Research Ecosystem
On the same day, DICP signed agreements with iFLYTEK, Alibaba Cloud, Huawei, and others to establish the Intelligent Chemical Industry Joint Innovation Lab, focusing on model and agent development, industry data and computing platforms, and industrial software integration.
The Intelligent Chemical Innovation Center, the first AI innovation platform for the chemical industry, was also established in Dalian. The team plans to continue developing version 4.0 with deep reasoning capabilities, using mature processes like methanol-to-olefins as validation scenarios to connect lab R&D, engineering design, and plant operations.
The real value of 3.0 Pro lies in answering a critical question for the chemical industry: AI must not only "understand" chemistry but also "do" chemistry. With validation from over 300 organizations and the establishment of recognized full-chain data infrastructure, this domestic AI foundation is moving from the lab to real production scenarios. When large models begin to understand catalytic mechanisms, invoke professional tools, and plan experimental workflows, the paradigm of chemical R&D is being redefined—a compelling example of new quality productive forces at work in a traditional industry.
Link：https://chemellm.dicp.ac.cn/