Truman Hickok’s paper introduces SO-101 Bench, an Isaac Lab digital twin of a physical SO-101 setup for language-conditioned spatial manipulation. It includes four tasks, 138 objects across seen, unseen-instance and unseen-class splits, 916 evaluation episodes and 3,629 simulated demonstrations; the digital twin reproduced the physical arm’s policy ranking with Spearman rho of 1.00, although simulation overestimated success. Across nine VLAs and world-action models, the best learned policies scored nearly 61%, while an LLM agent combining code generation with a learned grasping policy reached 83.2%.