We engineered a next-generation vision-language model (VLM) to handle complex spatial reasoning tasks; we call it Spatial Intelligence. Its goal is to give users the ability to reliably accomplish tasks remotely.

Unlike today’s VLMs which only answer basic questions, Spatial Intelligence accurately handles complex tasks like estimating distances, comparing object sizes, or understanding how objects are arranged in space.

Spatial Intelligence builds on our TeleForm model. It can use all the information in a scene, has temporal context, and applies chain-of-thought reasoning to solve complex problems.

We’re excited to share a limited preview of this technology before its full release next year.