Intern-S1-Pro: Scientific Multimodal Foundation Model at Trillion Scale

TL;DR

Intern-S1-Pro is a groundbreaking one-trillion-parameter scientific multimodal foundation model that excels in general reasoning, scientific tasks, and agent capabilities across multiple scientific domains.

Contribution

The paper introduces the first trillion-parameter scientific multimodal foundation model with extensive scientific expertise and efficient RL training infrastructure.

Findings

Outperforms proprietary models in scientific tasks

Mastered over 100 specialized science tasks

Demonstrates strong general and scientific reasoning capabilities

Abstract

We introduce Intern-S1-Pro, the first one-trillion-parameter scientific multimodal foundation model. Scaling to this unprecedented size, the model delivers a comprehensive enhancement across both general and scientific domains. Beyond stronger reasoning and image-text understanding capabilities, its intelligence is augmented with advanced agent capabilities. Simultaneously, its scientific expertise has been vastly expanded to master over 100 specialized tasks across critical science fields, including chemistry, materials, life sciences, and earth sciences. Achieving this massive scale is made possible by the robust infrastructure support of XTuner and LMDeploy, which facilitates highly efficient Reinforcement Learning (RL) training at the 1-trillion parameter level while ensuring strict precision consistency between training and inference. By seamlessly integrating these advancements, Intern-S1-Pro further fortifies the fusion of general and specialized intelligence, working as a Specializable Generalist, demonstrating its position in the top tier of open-source models for general capabilities, while outperforming proprietary models in the depth of specialized scientific tasks.