SenseNova-SI - A Family of Open Source Spatial Intelligence Large Models from ShangTech

What is SenseNova-SI

SenseNova-SI is an open source spatial intelligence grand model released by ShangTech, focusing on improving AI's ability in spatial understanding and reasoning. The model excels in six core dimensions, including spatial measurement, reconstruction, relationship judgment, perspective transformation, deformation analysis, and spatial reasoning, significantly outperforming other open-source and closed-source models. For example, in complex road scenes, SenseNova-SI can accurately judge the follow-up actions of vehicles, which is difficult for other models to do, and SenseNova-SI adopts a systematic training method to validate the "scale effect" through large-scale high-quality data training, which significantly improves spatial cognitive ability. Based on the multimodal basic model construction, through continuous training, the model has stronger spatial understanding ability in complex scenes.

SenseNova-SI functional features

• Spatial Measurement: It can accurately measure the size, distance and other spatial parameters of objects, providing basic data support for scene analysis.
• spatial reconstruction: A complete 3D spatial structure can be reconstructed based on partial information, which is suitable for modeling and understanding complex scenes.
• Judgment of spatial relations: Accurately recognizes the relative positions and spatial relationships between objects, such as up and down, front and back, and containment, to enhance the depth of scene understanding.
• perspective shift: Supporting the understanding and analysis of the same scene from different viewpoints, enhancing the model's adaptability to multi-view data.
• Spatial deformation analysis: It can recognize and predict the deformation of objects in space, such as stretching and bending, and is suitable for the analysis of dynamic scenes.
• spatial reasoning: With strong spatial logic reasoning capability, it can infer unknown spatial relationships or events based on known information, enhancing the decision-making capability of the model.

Core Benefits of SenseNova-SI

• Spatial intelligence leads the way: Averaged 60.99 points in 6 authoritative reviews of space missions, significantly outperforming closed-source models such as GPT-5 and Gemini 2.5 Pro and mainstream open-source models.
• Open source dual specification: 2B/8B parameter model all open source, 8B version of the industry's best, developers zero threshold for commercial use.
• The first "law of scale" for space intelligence.: Validating that massive high-quality data can systematically improve spatial cognition and lay the methodology for subsequent iterations.
• Synchronized open source EASI evaluation platform: Provide unified benchmarks and public lists to drive the establishment of industry standards for spatial intelligence.
• Direct empowerment with embodied intelligence: Accurate spatial measurement, reconstruction and inference to provide centimeter-level understanding for applications such as autonomous driving and robot navigation.

What is SenseNova-SI's official website?

• GitHub repository:: https://github.com/OpenSenseNova/SenseNova-SI
• HuggingFace Model Library:: https://huggingface.co/collections/sensenova/sensenova-si

Who is SenseNova-SI for?

• Artificial intelligence researchers: It can be used to explore cutting-edge technologies in spatial intelligence and to advance academic research and model optimization.
• Autonomous Driving Technology Developers: Provide accurate spatial measurement and reasoning capabilities to aid scene understanding and decision making in autonomous driving systems.
• Robotics Engineer: Help robots better understand 3D space and improve navigation, obstacle avoidance, and task execution.
• game developer: for building more realistic and intelligent game scenarios and character behavior.
• Industrial Automation Specialist: For spatial analysis and automated process optimization in industrial environments.
• Educators and students: as a teaching tool to help students understand spatial intelligence and multimodal technology.
• Research and development staff of science and technology enterprises: Can be used to develop innovative products and services based on spatial intelligence.