RoboSPA Benchmark
Explore score history ↗RoboSPA is a large-scale robotic manipulation benchmark that evaluates Vision-Language-Action models on fine-grained spatial reasoning and long-horizon procedural planning. It includes 10 task categories and 56 base tasks, each instantiated at five difficulty levels to yield 280 variants, with 527K trajectories. Beyond binary success, it reports diagnostic metrics such as ONTA and Progress Score. Code and data are available.
- Released
- 2026-09-04
- Evaluates
- Reasoning, Robot manipulation, Robotics, Robotics & Embodied AI
- Openness
- unknown
- Importer
- Claire Radar
- Review status
- ai-name-audit-deferred
RoboSPA paper and code
No reported scores are on record for this benchmark yet.