Search papers, labs, and topics across Lattice.
S1-VL, a 32B parameter multimodal model, is introduced to enhance scientific reasoning by integrating structured chain-of-thought with active image manipulation via Python code execution. A novel six-dimensional quality filtering framework and adaptive data routing strategy were developed to curate a high-quality scientific multimodal dataset spanning six disciplines, addressing issues like redundant visual operations. Trained using a four-stage pipeline including supervised fine-tuning and reinforcement learning with SAPO, S1-VL achieves state-of-the-art performance on Thinking-with-Images benchmarks and demonstrates strong scientific reasoning abilities.
Scientific reasoning gets a visual upgrade: S1-VL lets models "think with images" by writing and executing Python code to manipulate visuals during multi-step problem solving.
We present S1-VL, a multimodal reasoning model for scientific domains that natively supports two complementary reasoning paradigms: Scientific Reasoning, which relies on structured chain-of-thought, and Thinking-with-Images, which enables the model to actively manipulate images through Python code execution during reasoning. In the Thinking-with-Images mode, the model generates and executes image-processing code in a sandbox environment, obtains intermediate visual results, and continues reasoning in a multi-turn iterative manner. This design is particularly effective for challenging scenarios such as high-resolution scientific chart interpretation, microscopic image understanding, and geometry-assisted reasoning. To construct the training data, we collect scientific multimodal datasets spanning six disciplines: mathematics, physics, chemistry, astronomy, geography, and biology. We further develop a six-dimensional quality filtering framework for reasoning trajectories. To mitigate redundant, ineffective, and erroneous visual operations commonly found in existing datasets, we propose a multi-stage filtering pipeline together with an adaptive data routing strategy. This strategy converts samples with low visual information gain into pure Reasoning-mode data, enabling the model to learn when image operations are truly necessary. S1-VL is trained through a four-stage progressive pipeline: scientific multimodal SFT, Thinking-with-Images cold-start SFT, and two stages of reinforcement learning with SAPO. We build S1-VL-32B on top of Qwen3-VL-32B-Thinking and evaluate it on 13 benchmarks. Experimental results show that S1-VL-32B achieves state-of-the-art performance on all five Thinking-with-Images benchmarks, including HRBench-4K, HRBench-8K, MME-RealWorld-CN, MME-RealWorld-Lite, and V*, and outperforms compared systems on scientific reasoning benchmarks such as Physics and VRSBench.
