88Z AI, 28 May — X Square Robot has released the open-source embodied large model Wall-OSS-0.5, which supports direct deployment after pre-training. The model performed well in zero-shot tests across 17 real-world robot tasks, with scores exceeding 80 in four of them.
At a glance
- Wall-OSS-0.5 supports direct deployment after pre-training, scoring over 80 in four zero-shot real-world tasks
- Pre-trained on diverse robot forms and massive data, it outperforms the industry benchmark π0.5 by 17.5 points after fine-tuning

The update
X Square Robot has officially open-sourced the embodied AI large model Wall-OSS-0.5, introducing the concept of "deployment ready after pre-training." In zero-shot tests across 17 real-world robot tasks, the model achieved scores above 80 in four tasks, demonstrating strong generalization capabilities. The model was pre-trained on data from over 20 robot forms, more than a million trajectory records, and 90 million multimodal corpus entries. Technically, it employs gradient bridge collaborative training and a vision-aligned action Tokenizer. After fine-tuning, its task progress performance significantly outperformed the industry benchmark π0.5 by 17.5 points, highlighting its technical advantages in the field of embodied intelligence.
Sources & editorial note
This article is an AI-assisted synthesis by 88Z AI based on the public sources below. Product claims and availability are attributed to their publishers. It does not represent independent testing or first-hand reporting.
- 机器之心Source published
Source dates above refer to the original announcements; the date under the headline is this article’s publication on 88Z AI.