2026
Intern-S2-Preview: Scientific Agentic Foundation Model
Lei Bai, Jiaqi Cao, ..., Zhouqi Hua, ..., Yicheng Zou et al.
Technical Report
Intern-S2-Preview is a 397-billion-parameter scientific agentic foundation model that advances multimodal scientific understanding, reasoning, generation, time-series modeling, and long-horizon tool use through scientific multimodal pre-training, scalable multi-task and agentic reinforcement learning, and on-policy distillation.
Intern-S2-Preview: Scientific Agentic Foundation Model
Lei Bai, Jiaqi Cao, ..., Zhouqi Hua, ..., Yicheng Zou et al.
Technical Report
Intern-S2-Preview is a 397-billion-parameter scientific agentic foundation model that advances multimodal scientific understanding, reasoning, generation, time-series modeling, and long-horizon tool use through scientific multimodal pre-training, scalable multi-task and agentic reinforcement learning, and on-policy distillation.
Intern-S1-Pro: Scientific Multimodal Foundation Model at Trillion Scale
Yicheng Zou, Dongsheng Zhu, ..., Zhouqi Hua, ..., Lei Bai et al.
Technical Report
Intern-S1-Pro is a one-trillion-parameter scientific multimodal foundation model that enhances general and scientific capabilities through advanced agent functionalities and specialized task mastery across multiple scientific disciplines.
Intern-S1-Pro: Scientific Multimodal Foundation Model at Trillion Scale
Yicheng Zou, Dongsheng Zhu, ..., Zhouqi Hua, ..., Lei Bai et al.
Technical Report
Intern-S1-Pro is a one-trillion-parameter scientific multimodal foundation model that enhances general and scientific capabilities through advanced agent functionalities and specialized task mastery across multiple scientific disciplines.
2025
Intern-S1: A Scientific Multimodal Foundation Model
Lei Bai, Zhongrui Cai, ..., Zhouqi Hua, ..., Yu Qiao et al.
Technical Report
Intern-S1 is a large multimodal MoE foundation model trained with massive scientific data and mixture-of-rewards reinforcement learning, achieving SOTA performance in scientific reasoning and professional tasks while remaining competitive in general reasoning among open-source models.
Intern-S1: A Scientific Multimodal Foundation Model
Lei Bai, Zhongrui Cai, ..., Zhouqi Hua, ..., Yu Qiao et al.
Technical Report
Intern-S1 is a large multimodal MoE foundation model trained with massive scientific data and mixture-of-rewards reinforcement learning, achieving SOTA performance in scientific reasoning and professional tasks while remaining competitive in general reasoning among open-source models.
The Imitation Game: Turing Machine Imitator is Length Generalizable Reasoner
Zhouqi Hua, Wenwei Zhang, Chengqi Lyu, Yuzhe Gu, Songyang Gao, Kuikun Liu, Dahua Lin, Kai Chen
International Conference on Learning Representations ICLR 2026
Turing Machine Imitation Learning (TAIL) is a synthetic chain-of-thought framework that instills Turing machine–like execution in LLMs, enabling robust length generalization for computable reasoning. On 18 challenging tasks, a 7B TAIL model outperforms the 671B DeepSeek-R1, establishing a new state of the art.
The Imitation Game: Turing Machine Imitator is Length Generalizable Reasoner
Zhouqi Hua, Wenwei Zhang, Chengqi Lyu, Yuzhe Gu, Songyang Gao, Kuikun Liu, Dahua Lin, Kai Chen
International Conference on Learning Representations ICLR 2026
Turing Machine Imitation Learning (TAIL) is a synthetic chain-of-thought framework that instills Turing machine–like execution in LLMs, enabling robust length generalization for computable reasoning. On 18 challenging tasks, a 7B TAIL model outperforms the 671B DeepSeek-R1, establishing a new state of the art.