Embodied First Exam benchmark finds coding agents struggle in physical world
Researchers including Fei-Fei Li released Embodied First Exam, a benchmark of 1,090 embodied tasks across 31 simulators and 44 robots, finding coding agents achieve only 34% success when controlling robots.
2026-11-17 ~ 2026-11-18 · 2 related posts
- Embodied First Exam: Li Fei-Fei et al. benchmark coding agents on robots, 34% top success — ChongZzZhang · 2026-11-17
- Embodied First Exam: 1,090 embodied tasks show agents crumble outside privileged settings — ManlingLi_ · 2026-11-18