Reportedly: SenseTime Open-Sources SenseNova-RoboRSI Embodied AI Harness and a 27B Robot Brain Model
Reported by Pandaily · not yet confirmed by the company or a second independent outlet. We update this page when it is.
SenseTime open-sourced the SenseNova-RoboRSI evaluation harness and a 27B embodied model, after a study of GPT-6-Astra across 12 simulated benchmarks and 8 real robots. Figures are SenseTime's own, Pandaily reported.
China context
- Original name
- 商汤科技
- Outside China
- Open weights · pandaily.com
- Claims
- Company-reported; not yet independently evaluated
- For builders
- Developers can access the SenseNova-RoboRSI harness and 27B model weights to benchmark and build embodied AI systems.
- For investors
- SenseTime's open-source move may signal a strategy to build an ecosystem around its embodied AI technology, but no commercial impact is yet evident.
SenseTime released the SenseNova-RoboRSI evaluation harness and a 27B embodied AI model as open source. The release follows a study of GPT-6-Astra across 12 simulated benchmarks and 8 real robots. All figures are reported by SenseTime.
The open-sourced harness and 27B model provide a reproducible evaluation setup for embodied AI. The study covered 12 simulated benchmarks and 8 real robots, but no independent verification of the results is available.
Robotics developers outside China gain access to a SenseTime evaluation harness and model weights, reducing the cost of benchmarking embodied AI systems. Competitors can now test against SenseTime's reported performance on 12 simulated and 8 real-robot tasks.
Open-sourcing the harness and model may attract developers to SenseTime's embodied AI stack, but no commercial terms or adoption data are provided.
Next signals to check: whether the model weights are actually downloadable from a public repository, and whether independent researchers reproduce the benchmark results.