Intel / post

We're releasing HY-Embodied-0.5, a family of foundation models for real-world…

Source: x.com
Date: Published Apr 10, 2026

Context

Tencent's HY-Embodied-0.5 is designed to help physical agents interpret scenes and reason about actions. Its release documentation describes transferring reasoning and planning behavior from a larger model into a compact variant intended for deployment near the device. The released MoT-2B model has roughly four billion total parameters and activates about 2.2 billion during . Its small-model designation therefore describes active computation, while the intended role is perception and reasoning within a broader robot-control system.

Terms in this piece · Glossary

inference — Running a trained model to get answers — the phase where AI is actually used, as opposed to trained.
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
token — The chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.
distillation — Training a small, cheap model to imitate a big one's outputs, keeping much of the capability at a fraction of the cost.

Read the source x.com