36Kr Publishes Article on Embodied Intelligence

36Kr reported that it published an article on embodied intelligence on November 13, 2026.

What can be confirmed here is that 36Kr published an article. We will not reconstruct the article’s argument or any figures or statements it contains. Presenting the issue addressed by the article is a separate step from verifying its content against primary sources or real-world evaluations. See the 36Kr article for details of the original. In our previous coverage of Physical Intelligence, we have distinguished reporting about models from verification of specifications and performance. humanoid.guide covers Physical Intelligence’s “π*0.6” and humanoid.guide reports on Physical Intelligence’s “π0.7” both report the publication of coverage about models, while noting that the information provided does not confirm official specifications or performance. The publication of this article should likewise not be confused with confirmation of a model’s performance or practicality. When evaluating technology, the key question is whether the model name and version, the tasks being evaluated, the conditions and the measurement methods are specified well enough to allow comparison. The simulation training and real-world application of humanoid control technology using world models covered by Nikkei xTECH should also not be treated as a single evaluation: results in simulation and reproducibility on real hardware need to be examined separately. This does not infer the content of the 36Kr article; it reflects points to keep in mind when assessing the coverage cited above. Next, it is necessary to check whether primary sources support the claims in the article and whether specific evaluation conditions have been disclosed. Publication alone does not establish the scale or extent of any data shortfall, or the effects on the training or operation of a particular model. Reporting that raises questions, technical specifications and real-world validation results need to be tracked separately.