X Square Robot open-sources Wall-OSS-0.5 VLA model

Evidence notes

• Wall-OSS-0.5 explores whether pretrained vision-language-action models can produce robot capabilities directly testable on physical hardware before task-specific fine-tuning.
• Technical highlights include gradient-bridged co-training, a Vision-Aligned RVQ Action Tokenizer, action-space supervision and the DMuon distributed optimizer.
• In zero-shot real-robot evaluation, the pretrained checkpoint achieved task-progress scores above 80 on Block Sorting, Fruit Sorting, Ring Stacking and Rope Tightening.